Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
in the simplest random - effects model of meta - analysis involving , say , @xmath0 studies the data is supposed to consist of treatment effect estimators @xmath1 , which have the form @xmath2 here @xmath3 is an unknown common mean , @xmath4 is zero mean between - study effect with variance @xmath5 , @xmath6 , and @xmath7 represents the measurement error of the @xmath8th study , with variance @xmath9 . then the variance of @xmath10 is @xmath11 . in practice @xmath12 is often treated as a given constant , @xmath13 , which is the reported standard error or uncertainty of the @xmath8th study . the considered here problem is that of estimation of the common mean @xmath3 and of the heterogeneity variance @xmath5 from the statistical decision theory point of view under normality assumption . if @xmath5 is known , then the best unbiased estimator of @xmath3 is the weighted means statistic , @xmath14 , with the normalized weights , @xmath15 @xmath16 . its variance has the form @xmath17 ^{-1}.\ ] ] when @xmath5 is unknown , to estimate @xmath3 the common practice uses a plug - in version of @xmath18 , @xmath19 so that an estimator @xmath20 of @xmath5 is required in the first place . usually such an estimator is obtained from a moment - type equation @xcite . for example , the dersimonian laird @xcite estimator of @xmath5 is @xmath21 with @xmath22 denoting the graybill deal estimator of @xmath3 . the popular dersimonian laird @xmath3-estimator is obtained from ( [ es ] ) by using the positive part of @xmath23 . similarly the estimator of @xmath5 , @xmath24 leads to the hedges estimator of @xmath3 . the paper questions the wisdom of using under all circumstances the tradition of plugging in @xmath5 estimators to get @xmath3 estimators . indeed the routine of plug - in estimators may lead to poor procedures . for example , by replacing the unknown @xmath5 by @xmath25 in the above formula for @xmath26 , one can get a flagrantly biased estimator which leads to inadequate confidence intervals for @xmath3 . a large class of weighted means statistics is motivated by the form of bayes procedures derived in section [ qe ] . these statistics which typically _ do not _ admit the representation ( [ es ] ) induce estimators of the weights ( [ we ] ) which shows the primary role of @xmath3-estimation . the main results of this work are based on a canonical representation of the restricted likelihood function in terms of independent normal random variables and possibly of some @xmath27-random variables . an important relationship between the weighted means statistics with weights of the form ( [ we ] ) and linear combinations of @xmath28 s , which are shift invariant and independent , follows from this fact . our representation transforms the original problem to that of estimating curve - confined expected values of independent heterogeneous @xmath29-random variables . this reduction makes it possible to describe the risk behavior of the weighted means statistics whose weights are determined by a quadratic form . we make use of the concept of permissible estimators which can not be uniformly improved in terms of the differential inequality in section [ inad ] . this inequality shows that the sample mean exhibits the stein - type phenomenon being an inadmissible estimator of @xmath3 under the quadratic loss when @xmath30 . a risk function for the weights in a weighted means statistic whose main purpose is @xmath3-estimation is suggested in section [ rrisk ] . it is shown there that under this risk the sample mean is not even minimax . section [ mini ] discusses the case of approximately equal uncertainties , and section [ exa ] gives an example . the derivation of the canonical representation of the likelihood function is given in the ; the proof of theorem [ th1 ] is delegated to the electronic supplement @xcite . the setting with the common mean @xmath3 and the heterogeneity variance @xmath5 described in section [ in ] is a special case of a mixed linear model where statistical inference is commonly based on the restricted ( residual ) likelihood function . the ( negative ) restricted log - likelihood function ( @xcite , section 6.6 ) has the form @xmath31.\ ] ] it is possible that some of @xmath32 are equal ; let @xmath32 have the multiplicity @xmath33 , so that @xmath34 . then with the index @xmath8 now taking values from @xmath35 to @xmath36 , @xmath37.\ ] ] here , @xmath38 denotes the number of pairwise different @xmath32 , @xmath39 represents the average of @xmath40 @xmath28 s corresponding to the particular @xmath41 , and @xmath42 is their sample variance when @xmath43 . to simplify the notation , we write @xmath10 for @xmath44 , so that @xmath45 . in our problem @xmath46 and @xmath47 , @xmath48 , form a sufficient statistic for @xmath3 and @xmath5 . throughout this paper , we assume that @xmath49 . otherwise all @xmath3-estimators reduce to the sample mean ( but see section [ mini ] where @xmath5-estimation for equal uncertainties is considered ) . the results in the relate the likelihood function @xmath50 to the joint density of @xmath51 independent normal , zero mean random variables @xmath52 . the @xmath53-dimensional normal random vector @xmath54 which is a linear transform of @xmath28 has zero mean ( no matter what @xmath3 is ) and the covariance matrix , @xmath55 , with @xmath56 larger than @xmath57 . to find these numbers , we introduce the polynomial @xmath58 of degree @xmath0 , and its minimal annihilating polynomial @xmath59 which has degree @xmath38 . define @xmath60 thus @xmath61 is a polynomial of degree @xmath51 which has only real ( negative ) roots , denoted by @xmath62 ( coinciding with the roots of @xmath63 different from @xmath64 ) . thus @xmath65 . note that @xmath66 . when @xmath67 , @xmath68 , and @xmath69 . according to ( [ def ] ) , @xmath70 so that by using ( [ co7 ] ) one gets @xmath71 \\[-8pt ] & & \hphantom{\mathcal{l } = \frac{1}{2 } \biggl[}+\sum _ i \frac{(\nu_i-1 ) u_i^2}{\tau^2+s_i^2 } + \sum_i ( \nu_i-1)\log \bigl(\tau^2+s_i^2 \bigr ) + \log n \biggr ] . \nonumber\end{aligned}\ ] ] the representation ( [ rl ] ) of the restricted likelihood function very explicitly takes into account one degree of freedom used for estimating @xmath3 , as it corresponds to @xmath51 independent zero mean , normal random variables @xmath72 with variances @xmath73 . in addition , this likelihood includes independent @xmath42 , each being a multiple of a @xmath29-random variable with @xmath74 degrees of freedom . when @xmath48 , @xmath42 is an unbiased estimator of @xmath75 , @xmath76 . for @xmath77 , @xmath78 with probability one . according to the sufficiency principle , all statistical inference about @xmath5 involving the restricted likelihood can be based exclusively on @xmath79 and @xmath80 . their joint distribution forms a curved exponential family whose natural parameter is formed by @xmath81 ( and perhaps by some @xmath82 ) . evaluation of the restricted maximum likelihood estimator ( reml ) @xmath20 is considerably facilitated by employing @xmath83 and @xmath84 . indeed ( [ rl ] ) shows that this estimator can be determined by simple iterations as @xmath85 with @xmath23 as a good starting point , and truncation at zero if the iteration process converges to a negative number . thus , @xmath20 is related to a quadratic form whose coefficients are inversely proportional to the estimated variances of @xmath86 and of @xmath87 ( cf . @xcite , section 8) . the form of the likelihood function @xmath88 also motivates the moment - type equations based on general quadratic forms , @xmath89 with positive constants @xmath90 . the moment - type equation written in terms of random variables @xmath83 and @xmath84 is @xmath91 = \biggl [ \sum_j q_j + \sum _ i ( \nu_i-1)r_i\biggr ] \tau^2 + \sum_j q_j t_j^2 + \sum_i ( \nu_i-1)r_i s_i^2.\ ] ] then the estimator of @xmath5 by the method of moments is @xmath92 unless @xmath5 is large , the probability that @xmath93 takes negative values is non - negligible . non - negative statistics @xmath94 are used to get @xmath3-estimators of the form ( [ es ] ) . the representations of two traditional statistics in section [ in ] easily follow , @xmath95 and @xmath96 a different method - of - moments procedure suggested by paule and mandel @xcite is based on solving the equation , @xmath97 which has a unique positive solution , @xmath98 , provided that @xmath99 . if this inequality does not hold , @xmath100 . because of ( [ co7 ] ) , the equation can be rewritten in terms of @xmath101 s and @xmath102 s as @xmath103 this representation allows for an explicit form of @xmath104 in some cases . indeed , when @xmath105 , @xmath106 $ ] , which is also the reml . when @xmath107 , @xmath108 , @xmath109 we conclude this section by noticing that the widely used heterogeneity index @xmath110 ( @xcite , page 117 ) in terms of @xmath101 s and @xmath111 s takes the from @xmath112 let us look now at the generalized bayes estimator of @xmath3 when @xmath113 is a prior distribution for @xmath114 while @xmath3 has the uniform ( non - informative ) prior . under the quadratic loss this estimator has the form with @xmath88 given in ( [ rl ] ) @xmath115 thus @xmath116 is a weighted means statistic with normalized weights , @xmath117^{-1 } \exp\{-\mathcal{l}\ } \,\mathrm{d}\lambda(\tau^2)$ ] , @xmath118 , which are shift invariant , @xmath119 for any real @xmath120 . ( any function of @xmath52 is shift invariant . ) indeed the use of restricted likelihood is tantamount to the practice of weighted means statistics with invariant weights as @xmath3 estimators . @xcite , section 9.2 ) . formula ( [ rep ] ) in the gives @xmath121 with @xmath72 discussed in section [ ba ] . positive coefficients @xmath122 ( the diagonal elements of the diagonal matrix @xmath123 defined in lemma [ lem ] ) can be found from ( [ co6 ] ) or rather from ( [ co5 ] ) ; @xmath124 is the posterior mean of @xmath81 , @xmath125 with @xmath126 . thus positive @xmath124 is designed to estimate @xmath81 , @xmath127 , and as a function of @xmath128 , @xmath124 decreases . the inequalities , @xmath129 , and @xmath130 , are equivalent . if @xmath131 and the support of @xmath113 has at least two points , @xmath116 does not admit representation ( [ es ] ) which suggests a more general class of @xmath3-estimators . namely , we propose to use weighted means statistics @xmath132 with weights @xmath133 . the bayes weights belong to a smaller part of this polyhedron , namely to the convex hull of the vectors with coordinates @xmath134 for @xmath6 . if @xmath20 is an estimate of @xmath5 , the weights corresponding to ( [ es ] ) , @xmath135 lie on the boundary of this convex hull . a corner point , @xmath136 , of the convex hull always is an inner point of the polyhedron . thus the focus in this paper is on estimators @xmath116 of @xmath3 , which admit the representation , @xmath137 with @xmath138 and @xmath122 as defined above . the last term in the right - hand side of ( [ ne ] ) can be viewed as an arguably necessary heterogeneity correction to @xmath139 . notice that ( [ ne ] ) does not need an estimate of @xmath5 as a prerequisite . since @xmath124 is an approximation to @xmath81 , when @xmath140 , the form of the reml @xmath20 in section [ ba ] suggests such an estimator : @xmath141_+/\sum w_j^2 $ ] . if some of the multiplicities exceed one , an estimator of @xmath142 can be derived from @xmath124 by replacing @xmath143 by @xmath32 . according to ( [ rep ] ) , @xmath18 as well as @xmath139 , has the form ( [ ne ] ) . in fact , all traditional statistics ( [ es ] ) admit this representation . we look now at the quadratic risk behavior of @xmath3-estimators of the form ( [ ne ] ) . if @xmath144 is such an estimator with positive normalized weights @xmath145 which are shift invariant functions of @xmath146 , then it is unbiased . its variance does not depend on @xmath3 and can be written as @xmath147 ^{-1 } + e ( \delta-{\hat{\mu}_{\mathrm{opt}}})^2\ ] ] by independence of @xmath18 and @xmath148 . this and more general decompositions of the mean squared error are discussed by harville @xcite . the second term in the right - hand side of this identity is an important variance component which shows how well @xmath116 approximates the optimal but unavailable @xmath18 , and which relates our setting to the classical estimation problem of the multivariate @xmath53-dimensional normal mean . [ pro2.1 ] if the coefficients @xmath149 defining the estimator ( [ ne ] ) are piecewise differentiable in @xmath101 s , then @xmath150 where @xmath151 . when @xmath152 , @xmath139 is an inadmissible estimator of @xmath3 under the quadratic loss . the omitted proof of proposition [ pro2.1 ] is based on ( [ rep ] ) , ( [ co5 ] ) , and on familiar integration by parts technique . it demonstrates linkage of our situation to the differential inequality of a statistical estimation problem @xcite . namely , if for some vector @xmath153 , @xmath154 , then @xmath155 is an unbiased estimate of @xmath156 . therefore @xmath157 , @xmath158 , improves on @xmath159 , @xmath160 , as a @xmath153-estimator provided that for all values @xmath161 , @xmath162 following @xcite , let us call a ( piecewise differentiable ) vector function @xmath163 _ permissible _ if ( [ per ] ) does not have any solutions @xmath164 providing a strict inequality at some point . thus , @xmath165 is a permissible estimator of the vector normal mean @xmath153 if and only if the corresponding scalar @xmath3-estimator , @xmath166 , can not be improved upon in the sense of differential inequality ( [ per ] ) . since for @xmath152 , @xmath167 is not a permissible function , the sample mean @xmath139 is inadmissible in the original setting . indeed the left - hand side of ( [ per ] ) is negative for @xmath168 proving this statement . the differential operator in ( [ per ] ) does not involve @xmath102 s or @xmath169 s , but in our problem only functions @xmath170 such that @xmath171 and @xmath172 are of interest . since @xmath81 is positive and can not exceed @xmath173 , according to the first equality in proposition [ pro2.1 ] , @xmath124 can be improved by @xmath174 $ ] . the proof of theorem 1 in @xcite shows that any permissible @xmath124 in our situation is of the form @xmath175\ ] ] with some piecewise differentiable positive function @xmath176 . when @xmath140 and @xmath177 for a positive quadratic form @xmath178 , one gets @xmath179 $ ] . if there are multiplicities exceeding one , the quadratic form @xmath180 is to be replaced by @xmath181 . for example , the function , @xmath182 , leads to the estimator ( [ ne ] ) with @xmath183.\ ] ] the statistic @xmath184 , corresponding to @xmath185 , when @xmath140 is similar to the positive part of the stein estimator of the vector normal mean which improves over @xmath186 . however , in the meta - analysis context it is desirable having the coefficients @xmath187 of the same ordering as @xmath173 , and this condition may not hold for @xmath188 . as a matter of fact , despite doing better than @xmath189 or @xmath190 , the weights @xmath191 do not produce a good estimator of @xmath3 . the same is true for many other procedures ( [ wco ] ) satisfying condition ( [ xbar ] ) of theorem [ th1 ] in the next section . this theorem shows that if @xmath192 , @xmath139 is an inadmissible estimator of @xmath3 although the function @xmath167 is permissible then . according to ( [ qr ] ) the variance of estimator ( [ ne ] ) is completely determined by the term , @xmath194 , which can be interpreted as a cost of not knowing @xmath5 when estimating @xmath3 , or as a new risk of @xmath116 viewed as a procedure providing approximations to @xmath195 . more conveniently , with @xmath196 , define @xmath197 ^ 2 } { ( { \tau^2+s^2})/{n}- [ \sum_i { \nu_i}/ ( { \tau^2+s_i^2 } ) ] ^{-1}}\ ] ] to be the @xmath193-risk of @xmath116 . because of ( [ rep ] ) and ( [ ne ] ) , the ensuing random loss function has the form , @xmath198^{-1 } } = \frac{\sum_j ( w_j -{1}/({\tau^2+t_j^2 } ) ) ^2 b_j y_j^2}{\sum_j { b_j}/ ( { \tau^2+t_j^2 } ) } .\ ] ] this loss is invariant under a scale change of @xmath199 ( or of @xmath200 ) . for @xmath201 , @xmath202^{-1 } \sim\frac{\sum_i \nu_i(s_i^2- s^2)^2}{n^2 \tau^2 } , \ ] ] so that the normalizing factor in the definition of @xmath203 amplifies the error in approximating @xmath18 when @xmath5 is large . the results of this section show that for estimators @xmath116 satisfying conditions of the following theorem [ th1 ] , @xmath204 \bigl[r\bigl(\delta , \tau^{2}\bigr)-1\bigr ] = \frac{\tau^2+s^2}{n } + \mathrm{o } \biggl ( \frac{1 } { \tau^2 } \biggr)\ ] ] when @xmath205 . thus , @xmath206 is the dominating contribution to the variance of @xmath116 when @xmath5 is large . the @xmath193-risk is a useful tool for the comparison of estimators ( [ ne ] ) , as unlike the normalized quadratic risk , @xmath207 , it removes this linear in @xmath5 term . if @xmath208 with an invariant @xmath20 , then @xmath209 can be interpreted as a conventional risk of the estimator @xmath20 . however under this risk large values of @xmath20 are not penalized very much no matter what @xmath5 is . indeed @xmath20 is not designed to estimate @xmath5 itself , but rather @xmath210 estimates @xmath211 ( cf . @xcite , page 329 ) . when @xmath105 , the estimator @xmath139 , which corresponds to @xmath212 , is even admissible which of course can not happen for any unbounded loss function . this circumstance explains why an estimator @xmath20 may have a large quadratic risk , while the associated estimator @xmath213 in ( [ es ] ) has a small variance . that phenomenon is known to happen in the case of the dersimonian laird procedure @xcite . the estimator @xmath214 has a constant risk , @xmath215 , which raises the question of its @xmath193-minimaxity , i.e. , if @xmath216 . in contrast , for the graybill deal estimator , @xmath217 /[\sum b_j ( \tau^2 + t_j^2)^{-1}]$ ] , so that its @xmath193-risk , which vanishes when @xmath218 , grows quadratically in @xmath5 . the next result gives a large class of estimators with bounded @xmath193-risk improving on @xmath139 when @xmath219 . [ th1 ] under notation of section [ ba ] , let for @xmath220 , @xmath221 be a quadratic form with positive coefficients @xmath90 . if @xmath116 has the form ( [ ne ] ) such that for @xmath222 , then @xmath223\quad\quad\ \ \\ & \geq&\frac{2}{n-1 } , \nonumber\end{aligned}\ ] ] where independent standard normal @xmath224 are independent of @xmath225 . equal coefficients @xmath226 ( and only they ) provide the asymptotically optimal quadratic form . if @xmath227 , the optimal choice is @xmath228 . the sample mean @xmath139 is not @xmath193-minimax , any estimator ( [ ne ] ) with weights ( [ wco ] ) improves on it if @xmath229 \sum_j b_j q_j } { \max [ \max_j q_j^2 t_j^ 4 , \max_{i : \nu_i \geq2 } r_i^2 s_i^4 ] \sum_j b_j q_j^2}.\ ] ] theorem [ th1 ] shows that the traditional weights ( [ weig ] ) with @xmath230 are not asymptotically optimal unless the quadratic form @xmath231 coincides ( up to a positive factor ) with @xmath232 , and @xmath233 . only then ( [ isr ] ) is an equality . thus , the hedges estimator for which @xmath234 and @xmath235 , is not asymptotically optimal albeit its performance is the best when @xmath5 is large . for the mandel paule estimator from section [ ba ] , as well as for the reml , ( [ isr ] ) also holds with the same quadratic form and the same @xmath236 . the dersimonian laird estimator is defined by the quadratic form @xmath237 with @xmath238 . therefore , these three statistics are not optimal for large @xmath5 either . the case when @xmath105 was studied in @xcite . then @xmath139 is admissible ( so that it is automatically minimax under @xmath193 ) . any estimator ( [ ne ] ) has the form ( [ es ] ) with some @xmath239 , and its @xmath193-risk grows linearly in @xmath240 , @xmath241 for @xmath242 , as @xmath205 , @xmath243 ( see electronic supplement ) . by analogy with the stein phenomenon , admissibility of the sample mean when @xmath244 is expected . when @xmath220 , the minimax value , @xmath245 , ( which does not exceed one since @xmath246 ) can not be smaller than @xmath247 . indeed for any estimator @xmath116 , @xmath248 this fact can be proven by constructing a sequence of proper prior densities for @xmath5 such that the corresponding sequence of the bayes @xmath193-risks converges to @xmath247 . thus for large @xmath5 , the estimators ( [ ne ] ) with @xmath249 , @xmath233 , can not be improved upon . the most natural of these statistics , say , @xmath250 has the form ( [ ne ] ) with @xmath251 another modified hedges estimator , @xmath252 , has the form ( [ es ] ) with @xmath253_+$ ] and also is asymptotically optimal although in general its performance is worse than that of ( [ del ] ) . if @xmath254 , so that all @xmath143 and @xmath32 tend to @xmath255 , @xmath256 , @xmath257 ^ 2 \,\mathrm{d}g_{n+1 } \biggl ( \frac{v } { \tau^2+s^2 } \biggr).\ ] ] here and further @xmath258 is the distribution function of @xmath29-law with @xmath259 degrees of freedom . thus if @xmath260 , our problem is that of estimation of the reciprocal of the scale parameter @xmath261 under the restriction , @xmath262 . the `` data '' @xmath263 in this problem is @xmath29-distributed , @xmath264 , and the invariant loss function , @xmath265 , corresponds to the @xmath193-risk . then the minimax value , @xmath247 , is the same as in the non - restricted ( @xmath266 ) parameter case @xcite . as in unrestricted scale parameter estimation , the generalized prior , @xmath267 , @xmath262 , or @xmath268 , provides a least favorable distribution . see also @xcite for more general results . thus in meta - analysis problems with @xmath32 exhibiting little variation , the minimax value is expected to stay close to @xmath247 . indeed when @xmath269 , @xmath270 , @xmath271 the formula ( [ rar ] ) shows that the estimator ( [ del ] ) is minimax unlike @xmath252 for which @xmath272_+ /(n-3 ) + s^2 \}^{-1}$ ] . the dersimonian laird rule , @xmath273_+ /(n-1 ) + s^2 \ } ^{-1 } , \alpha = n-1 $ ] , coincides in this situation with the reml and the hedges estimator . for the proper maximum likelihood estimator of @xmath274 , @xmath275 . none of these procedures is minimax which indicates that their good properties in meta - analysis may be attributable to a large number of individual studies ( large @xmath0 ) or to lack of interest in high heterogeneity ( small @xmath5 ) . figure [ fi1 ] shows the graphs of the @xmath193-risk in ( [ rar ] ) when @xmath276 . it suggests that the estimator @xmath250 performs quite well for small / medium @xmath0 s . indeed this estimator is better than other procedures except for small @xmath5 in which case @xmath277 dominates @xmath252 ( at the price of higher risk for larger values of @xmath5 ) . -risks of estimators corresponding to @xmath278 ( dash - dotted line ) , @xmath252 ( line marked by diamonds ) and @xmath250 ( line marked by @xmath279 ) when @xmath280 ( left panel ) , and of the same risks when @xmath281 ( right panel ) . the straight line depicts the minimax value @xmath282 . ] when there are only two different values @xmath284 and @xmath285 with multiplicities @xmath286 and @xmath287 , @xmath288 , @xmath289 and if @xmath290 , @xmath291 any estimator ( [ ne ] ) has the form ( [ es ] ) for some @xmath20 , @xmath292 for @xmath250 , @xmath293_+ , q^\infty = y^2+(\nu_1 - 1)u_1 ^ 2+(\nu_2 - 1)u_2 ^ 2 $ ] . the modified hedges estimator @xmath252 with @xmath294_+$ ] typically has its @xmath193-risk at @xmath218 larger than that of @xmath250 . ( the exact condition for @xmath252 to have a smaller @xmath193-risk at @xmath218 than @xmath295 is : @xmath296 , and if @xmath297 , then @xmath298 , @xmath299 s_2 ^ 2 \geq [ n(n-4)-\nu_1(2n-5 ) ] s_1 ^ 2 $ ] . ) the @xmath193-risk of @xmath250 at @xmath218 can be larger than @xmath300 . indeed @xmath301 where @xmath302 is the distribution function of @xmath303 . with @xmath304^{1/(n+1)}$ ] , according to the okamoto inequality @xcite , @xmath305 . thus , since @xmath306 ^ 2 $ ] is an increasing function of @xmath307 , @xmath308}{n-1 } \\ & & { } + \frac { ( n-3)a^2[1-g_{n-3}(a(n-3))]}{n-1}.\end{aligned}\ ] ] this inequality shows that @xmath309 , if @xmath310 , where @xmath311 is monotonically increasing to @xmath35 in @xmath312 for small @xmath313 , @xmath250 can not have its risk at the origin smaller than @xmath282 . for example , when @xmath314 , @xmath315 if and only if @xmath316 , i.e. , iff @xmath317 . the dersimonian laird estimator @xmath277 with @xmath318 , has its @xmath193-risk at @xmath218 of the form @xmath319 ^ 2 \,\mathrm{d}g_{n+1}(v)\ ] ] with @xmath320 $ ] . for the estimator @xmath321 defined by ( [ ne ] ) , @xmath322 so that @xmath323_+$ ] . its risk at @xmath218 , @xmath324 is always smaller than that of @xmath250 . but @xmath321 is also competitive against @xmath277 . indeed @xmath325 if and only if @xmath326 , that is , iff @xmath327 thus provided that @xmath328}$ ] , @xmath329 improves upon @xmath277 for small @xmath5 . if , say , @xmath330 , this domination means that @xmath331 . thus , when one study reports a smaller uncertainty than all other studies whose standard errors are approximately equal , @xmath321 improves upon the dersimonian laird estimator for small @xmath5 . however , there is no uniform domination as the condition , @xmath332 , means that for large @xmath333 . author s attempt was to give a perspective of a meta - analysis setting from the point of view of the statistical decision theory . although concepts like admissibility or minimaxity have not so far generated much interest among meta - analysts , there is a realization that different desirable qualities of the employed procedures call for different loss functions . the quadratic loss for the mean effect estimators from a wide class leads in a natural way to the @xmath193-risk suggested and studied in this paper . this risk strongly recommends against the use of the sample mean as a consensus estimate which still happens in some collaborative studies . moreover , the @xmath193-risk questions well recognized excellent properties of the dersimonian laird estimator @xmath277 in the situation when @xmath13 are almost equal , or when one study claims a high precision while all other studies report larger uncertainties which are about the same . the unsatisfactory performance of the graybill deal estimator @xmath334 is well known in the latter case . it is of interest that @xmath321 improves on the dersimonian laird estimator for moderate / small @xmath5 . inference on the overall effect can be obtained before the heterogeneity variance is estimated , but even in the simplest cases considered here there is no unique rule dominating all others . this paper is dedicated to the memory of george casella who was always interested in implications of the statistical decision theory results to practical estimation problems @xcite . let @xmath335 denote unit coordinates vector whose dimension is clear from the context , and put @xmath336 , @xmath337 . in the used here notation of section [ ba ] the vector @xmath28 has the diagonal covariance matrix , @xmath338 . [ lem ] for any @xmath263 different from @xmath339 , and for any @xmath340 , @xmath341 ^{-1}= \frac{\nu_i } { n } -\sum _ { j } \frac { a_{ij } } { v+t_j^2},\ ] ] where @xmath342 for any @xmath343 , @xmath344 if the @xmath345 matrix @xmath346 is determined by its elements @xmath347 in ( [ coe ] ) , then @xmath348 and @xmath349 the matrices @xmath350 and @xmath351 are diagonal , and @xmath352 with @xmath353 , @xmath354 and @xmath355 by the definition of the polynomial @xmath61 in section [ ba ] , @xmath356 ^{-1 } & = & \frac{\nu_i } { n } - \frac{\nu_i p(v ) } { ( v+s_i^2 ) p^\prime(v ) } \\ & = & \frac{\nu_i } { n } -\frac{\nu_i m(v ) } { ( v+s_i^2 ) q ( v)}= \frac { \nu_i [ \prod_j ( v+t_j^2 ) - \prod_{k \neq i } ( v+s_k^2 ) ] } { q(v)}\end{aligned}\ ] ] with the right - hand side of this identity being the ratio of two polynomials of degree @xmath357 and @xmath51 , respectively . the formulas ( [ coo ] ) and ( [ coe ] ) easily follow from the classical result on partial fraction decomposition for such ratios . by equating coefficients at @xmath361 of @xmath362 and @xmath363 , one gets @xmath364 . the comparison of coefficients at @xmath365 of two equal polynomials , @xmath366 and @xmath367 $ ] , shows that @xmath368 which implies ( [ co2 ] ) . to prove ( [ co3 ] ) , observe that for @xmath377 , the @xmath378th element of the matrix @xmath379 has the form , @xmath380=- \frac{\nu_i \nu _ k}{n}.\end{aligned}\ ] ] here we used the facts that @xmath381 , and @xmath382 . to prove ( [ co10 ] ) for fixed @xmath386 , multiply ( [ coo ] ) by @xmath387 , divide by @xmath388 , and sum up over @xmath8 to get the following expression for the @xmath389th element of the matrix @xmath390 , @xmath391,\ ] ] where @xmath392 is the kronecker symbol ( @xmath393 , if @xmath394 otherwise ) . it is easy to see that @xmath395 , unless there are at least two equal indices among @xmath396 . when all three of these indices coincide , @xmath397 ^ 3 } \sum_i \frac{\nu_i } { ( s_i^2-t_j^2)^3 } \\ & = & -\frac { m(-t_j^2 ) [ q^{\prime\prime } ( -t_j^2 ) m(-t_j^2)- 2 q^\prime ( -t_j^2 ) m^\prime ( -t_j^2 ) ] } { 2 [ q^\prime ( -t_j^2)]^3 } \\ & = & \frac { b_j [ q^{\prime\prime } ( -t_j^2 ) m(-t_j^2)- 2 q^\prime ( -t_j^2 ) m^\prime ( -t_j^2 ) ] } { 2 [ q^\prime ( -t_j^2)]^2 } = - \frac { b_j^2 q^{\prime\prime } ( -t_j^2)+ 2 b_j m^\prime ( -t_j^2)}{2 q^\prime ( -t_j^2)}.\end{aligned}\ ] ] if , say , @xmath398 , @xmath399 ^ 2 q^{\prime}(-t_\ell^2 ) } \sum_i \frac{\nu _ i } { ( s_i^2-t_j^2)^2 ( s_i^2-t_\ell^2 ) } \\ & = & - \frac { m^2(-t_j^2 ) m(-t_\ell^2 ) } { [ q^{\prime } ( -t_j^2)]^2 q^{\prime}(-t_\ell^2 ) ( t_j^2-t_\ell^2 ) } \sum_i \frac{\nu_i } { ( s_i^2-t_j^2)^2 } \\ & = & \frac { m(-t_j^2 ) m(-t_\ell^2 ) } { q^{\prime } ( -t_j^2 ) q^{\prime } ( -t_\ell^2 ) ( t_j^2-t_\ell^2 ) } = \frac { b_j b_\ell}{t_j^2-t_\ell^2}.\end{aligned}\ ] ] the last formula shows that off - diagonal elements of @xmath400 , for @xmath401 have the form @xmath402 \\ & & \quad= \biggl(\sum_i \frac{\nu_i}{\tau^2+s_i^2 } \biggr)\frac{b_j b_\ell } { ( \tau^2 + t_j^2 ) ( \tau^2 + t_\ell^2)},\end{aligned}\ ] ] that is , ( [ co10 ] ) holds for the off - diagonal elements . define the polynomial @xmath405 by the formula , @xmath406 then the degree of @xmath407 is @xmath357 , and this polynomial is determined by its values at @xmath408 : @xmath409 , and @xmath410 . it follows that @xmath411 indeed , the polynomial in the right - hand side has degree @xmath357 . since @xmath412 it assumes the same values as @xmath407 at @xmath413 , which establishes ( [ co10 ] ) . because of ( [ co3 ] ) and ( [ coe ] ) , @xmath414x + ( { \rho}^{\mathrm{t } } a^{\mathrm{t } } x ) e $ ] . thus the quadratic form in the left - hand side of ( [ co7 ] ) can be written as @xmath415^{\mathrm{t } } c^{-1 } \bigl[j^{-1}a \bigl(a^{\mathrm{t } } j^{-1}a\bigr)^{-1 } a^{\mathrm{t } } x+ e { \rho}^{\mathrm{t } } a^{\mathrm{t } } x\bigr ] \\ & & \quad = y^{\mathrm{t } } \bigl [ \bigl(a^{\mathrm{t } } j^{-1}a \bigr)^{-1/2 } a^{\mathrm{t } } j^{-1}+\bigl(a^{\mathrm{t } } j^{-1}a\bigr)^{1/2 } { \rho}e ^{\mathrm{t}}\bigr ] \\ & & \qquad{}\times c^{-1}\bigl [ j^{-1}a \bigl(a^{\mathrm{t } } j^{-1}a \bigr)^{-1/2 } + e { \rho}^{\mathrm{t } } \bigl(a^{\mathrm{t } } j^{-1}a\bigr)^{1/2 } \bigr]y \\ & & \quad= y^{\mathrm{t } } \operatorname{diag}({\rho})y,\end{aligned}\ ] ] where the second equality follows from ( [ co10 ] ) and ( [ co9 ] ) . the following important representation for @xmath18 @xmath416 is a consequence of lemma [ lem ] . here @xmath417 are independent normal , zero mean random variables with the variances @xmath418 . indeed the normal random vector @xmath419 has the covariance matrix @xmath420 . since @xmath421 , @xmath18 and @xmath72 are independent implying independence of @xmath18 and @xmath148 in section [ inad ] . the coefficients @xmath347 provide a simple expression for @xmath422 . indeed , by dividing ( [ coo ] ) by @xmath40 and multiplying it by @xmath423 , one gets after summing up over all @xmath8 and @xmath371 and using ( [ co0 ] ) , ( [ co2 ] ) , @xmath424 ^{-1 } -\frac { \tau^2+s^2}{n } = -\sum_{i , j } \frac{a_{ij}^2}{\nu_i ( \tau^2 + t_j^2)}.\end{aligned}\ ] ] this formula can be written in the form , @xmath425 which provides the representation of the left - hand side of ( [ co5 ] ) as a ratio of two polynomials of degree @xmath357 and @xmath51 , respectively and which allows numerical evaluation of @xmath426 s without calculating @xmath347 .	in the random - effects model of meta - analysis a canonical representation of the restricted likelihood function is obtained . this representation relates the mean effect and the heterogeneity variance estimation problems . an explicit form of the variance of weighted means statistics determined by means of a quadratic form is found . the behavior of the mean squared error for large heterogeneity variance is elucidated . it is noted that the sample mean is not admissible nor minimax under a natural risk function for the number of studies exceeding three .
the branching fractions for the leptonic decays @xmath9 ( @xmath10 ) and @xmath11 ( @xmath12 ) are basic parameters of the @xmath3 resonance . they can be used to determine the strong coupling constant @xmath8 or , equivalently , the fundamental scale parameter of qcd , @xmath13 @xcite . the ratio @xmath14 provides a test of lepton universality . in addition , these branching fractions are used to determine the total number of @xmath3 events in a wide variety of measurements that take advantage of the clean experimental @xmath15 @xmath16 or @xmath17 signature . the first reported measurement of the leptonic @xmath3 branching fractions has a precision of about 15% and is based on an energy scan across the resonance performed by the mark - i group @xcite . a subsequent measurement by the mark - iii group @xcite is based on a comparison of the rates for @xmath0 , @xmath18 and @xmath2 anything and is , thus , independent of the luminosity determination . the mark - iii measurement has a precision of 4% and is about one ( mark i ) standard deviation below the mark - i result . more recently , bes performed an energy scan measurement and obtained results in good agreement with mark - iii , but with larger errors @xcite . in this paper , we report the results of a measurement of the leptonic branching fractions using a sample of @xmath19 decays measured in the bes detector at the bepc storage ring . we apply a technique similar to that used by the mark - iii group to a larger data sample . the beijing electron spectrometer , bes , is a conventional cylindrical magnetic detector that is coaxial with the bepc colliding @xmath20 beams . it is described in detail in ref . a four - layer central drift chamber ( cdc ) surrounding the beampipe provides trigger information . outside the cdc , the forty - layer main drift chamber ( mdc ) provides tracking and energy - loss ( @xmath21 ) information on charged tracks over @xmath22 of the total solid angle . the momentum resolution is @xmath23 ( @xmath24 in gev / c ) , and the @xmath21 resolution for hadron tracks for this data sample is @xmath25 . an array of 48 scintillation counters surrounding the mdc provides measurements of the time - of - flight ( tof ) of charged tracks with a resolution of @xmath26 ps for hadrons . outside the tof system , a 12 radiation length lead - gas barrel shower counter ( bsc ) , operating in self - quenching streamer mode , measures the energies of electrons and photons over 80% of the total solid angle . the energy resolution is @xmath27 ( @xmath28 in gev ) . surrounding the bsc is a solenoidal magnet that provides a 0.4 tesla magnetic field in the central tracking region of the detector . three double layers of proportional chambers instrument the magnet flux return ( muid ) and are used to identify muons of momentum greater than 0.5 gev / c . endcap time - of - flight and shower counters extend coverage to the forward and backward regions . our measurement is based on a data sample corresponding to an integrated luminosity of about 6.1 pb@xmath29 accumulated at the @xmath19 resonance . the @xmath19 is a copious source of @xmath3 decays : the branching fraction @xmath30 @xcite , is the largest single @xmath19 decay channel . we determine the @xmath3 leptonic branching fraction from a comparison of the exclusive and inclusive processes : = @xmath31 @xmath32=@xmath33 + @xmath34= @xmath35 + and @xmath36 anything @xmath37 + the @xmath3 leptonic branching fraction is determined from the relation @xmath38 where @xmath39 and @xmath40 are observed numbers of events for processes i and ii , and @xmath41 and @xmath42 are the respective acceptances . for both processes i and ii , we use only runs of good quality . we require at least one pair of oppositely charged candidate pion tracks that each satisfy the following criteria : 1 . @xmath43 gev / c , where @xmath44 is the pion momentum . @xmath45 gev / c , where @xmath46 is the momentum of the pion transverse to the beam direction . this removes tracks that circle in the main drift chamber . 3 . @xmath47 . here @xmath48 is the polar angle of the @xmath49 in the laboratory system . 4 . @xmath50 . @xmath51 is the laboratory angle between the @xmath52 and @xmath53 . this cut is used to eliminate contamination from misidentified @xmath20 pairs from @xmath54 conversions , as shown in fig . [ fig : cospipi ] . + 3.0 truein for leptonic decay candidate events ( process - i ) , the number of charged tracks is required to be at least four with a 4-track combination of net charge zero @xcite . lepton pair candidates must satisfy the following selection criteria : 1 . @xmath59 gev / c . here @xmath60 is the three - momenta of the candidate lepton track . @xmath61 gev / c or @xmath62 gev / c or @xmath63 gev / c . this cut selects events consistent with @xmath3 decay , while rejecting background . 3 . @xmath64 , @xmath65 . here @xmath66 and @xmath67 are the polar angles of the electron and muon , respectively . this cut ensures that electrons are contained in the @xmath68 and muons in the muid system . @xmath69 , where @xmath70 is the angle between the two leptons in the rest frame of the @xmath3 . 5 . for @xmath20 candidate pairs : @xmath71 and @xmath72 gev / c , where @xmath73 is the energy deposited in the bsc , or , if one of the tracks goes through a bsc rib or has @xmath74 gev / c , the @xmath21 information of both tracks in the mdc must be consistent with that expected for electrons . the rib region of the bsc is not used because the monte carlo does not model the energy deposition well in this region . 6 . for @xmath75 pair candidates at least one track must have @xmath76 , where @xmath77 is the number of muid layers with matched hits and ranges from 0 to 3 . if only one track is identified in this fashion , then the invariant mass of the @xmath78 pair must also be within 250 mev / c@xmath79 of the @xmath3 mass . the number of process - ii events ( @xmath2 anything ) is determined from a fit to the @xmath82 recoil mass spectrum , using a @xmath3 line shape that is determined from the recoil mass spectrum for the leptonic decays ( fig . this is fit with a breit - wigner function folded with a gaussian ( bwg ) and a third order background polynomial @xcite . the parameters thus obtained for the bwg fit are then used together with a fourth order polynomial background to fit the @xmath82 recoil mass for the inclusive @xmath3 decays , as shown in fig . the resulting number of inclusive decays is @xmath83 , where the error is the statistical uncertainty combined with the uncertainty in the fitting procedure . the same bwg parameters and a third order background are used to obtain the number of @xmath84 and @xmath85 events separately and yield a total of @xmath86 @xmath87 events and @xmath88 @xmath89 events . the results are summarized in table [ summary ] . the experimental ratio of the quarkonium annihilation rates @xmath90 allows the determination of the strong coupling constant @xmath8 or , equivalently , the qcd scale parameter @xmath91 @xcite . following the notation of ref . @xcite , one obtains @xmath92/ b(j/\psi\rightarrow l^{+}l^{-})\end{aligned}\ ] ] where , @xmath93 @xmath94 and @xmath95 the factor @xmath96 is the reduction factor of the three gluon decay caused by the finite size effect in the matrix element . for @xmath97 , we use the theoretically calculated rather than the measured value because of the large error associated with the latter @xcite . combining the above @xmath3 decay parameters and the values for @xmath98 and @xmath99 listed in pdg@xcite , we obtain @xmath100 @xmath101 where only the experimental errors are included . our results are in good agreement with the pdg@xcite values . work supported in part by the national natural science foundation of china under contract no . 19290400 and the chinese academy of sciences under contract no . kj85 ( ihep ) , and by the department of energy under contract nos . de - fg03 - 92er40701 ( caltech ) , de - fg03 - 93er40788 ( colorado state university ) , de - ac03 - 76sf00515 ( slac ) , de - fg03 - 91er40679 ( uc irvine ) , de - fg03 - 94er40833 ( u hawaii ) , de - fg03 - 95er40925 ( ut dallas ) . approximately 4% of the events have more than four prongs . these are usually due to very low energy @xmath49 track being reconstructed into two tracks . the monte carlo does not model this effect well , so it is necessary to include these events , even though the @xmath102 distribution is not as clean for these events . we rely on our fitting procedure to extract the signal . the events represented by the polynomial in fig 2a are predominantly good events but where the low energy pion underwent @xmath103 decay . using only the bwg shape to describe the signal makes us less sensitive to background . for process - ii , @xmath105 , the total number of events observed with charged multiplicity @xmath106 is related to the @xmath107 , the number generated with charged multiplicity @xmath108 , by the relation @xmath109 where the matrix element @xmath111 is the probability of an observed @xmath106-@xmath112 event coming from a @xmath108-@xmath113 event . summing over @xmath106 and using @xmath114 we have @xmath115 where @xmath116 is the acceptance for and @xmath117 the fraction of multiplicity-@xmath108 events . by solving the overdetermined equations above , we estimated @xmath3 charged - track - multiplicity distribution shown in table [ accept ] . using variations like those in table [ tab : varoldandnew ] , we can determine the systematic errors associated with @xmath118 . we obtain @xmath119 . using the branching ratio of @xmath120 given by @xcite , the produced number of @xmath19 events corresponding to the data used in this analysis is estimated to be @xmath121 . other analyses may use , in addition , runs of fair quality . using good and fair quality runs , yields @xmath122 @xmath3 events and @xmath123 @xmath19 events .	a comparison of the rates for @xmath0 , @xmath1 and @xmath2 anything is used to determine the @xmath3 leptonic branching fractions . the results are @xmath4 and @xmath5 , where the first error is statistical and the second is systematic . assuming lepton universality , the leptonic branching fraction of the @xmath3 is @xmath6 per species . this result is used to estimate the qcd scale factor @xmath7 and the strong coupling constant @xmath8 .
supersonic flows observed as collimated jets are frequently associated with young protostars of different masses and evolutionary stages . jets represent an essential tool by which the driving source removes its excess angular momentum , allowing the collapse to go on . their interaction with the ambient medium occurs via excitation of atoms and ions , whose cooling lines represent the way through which jets are seen . such cooling lines spread over a wide frequency range from optical - uv up to far - ir . in particular , molecular hydrogen imaging of the @xmath4 = 1 - 0 s(1 ) line at 2.12 @xmath1 m is extensively used for identifying knots of molecular emission , while the spectroscopy of the h@xmath0 ro - vibrational lines lying in the near - ir range ( 1 - 5@xmath1 m ) is largely effective both to probe the molecular gas at thousands kelvin and to infer the main excitation mechanism(s ) ( fluorescence , c - ontinous or j - ump shocks , see e.g. black @xmath5 van dishoeck 1987 ; kaufman & neufeld 1996 ; hollenbach @xmath5 mckee 1989 ; smith 1995 ) . statistically , jet images are powerful in correlating the properties of the young population of the parental cloud ( e.g. age , degree of clustering , initial luminosity function - ilf ) with their common features ( e.g velocities , dynamical ages , collimation and position angles ) . such correlations are undoubtedly fundamental for investigating how the cloud parameters affect the onset , the properties and the morphology of the jets and how these latter influence the cloud itself . during the last two decades , hundreds of individual young stellar objects ( ysos ) have been imaged with near - infrared narrow band filters ( h@xmath0 and [ fe ii ] ) , although extensive surveys able to cover square degree scale star forming regions are not so numerous ( for some exceptions see eg . stanke et al . 1998 , davis et al . 2008 , 2009 ) . in this sense the digital catalog of molecular hydrogen objects ( mho , davis et al . 2010 ) undoubtedly represents a relevant tool of investigation . the _ infrared array camera _ ( irac , fazio et al . 2004 ) on board of the _ spitzer space telescope _ ( werner et al . 2004 ) offers a unique chance to carry out large field searches of both jets and their obscured exciting sources ( e.g. cyganowski et al . 2011 ) , indeed , irac band - passes ( between 3.6@xmath1 m and 8.0@xmath1 m ) contain numerous h@xmath0 emission lines ( see fig.1 of smith & rosen 2005 ) and are therefore suited to find jets and to study their properties , as testified by an increasing number of observational works ( e.g. teixeira et al . 2008 ; smith et al . 2006 ; velusamy et al . 2007 ; neufeld & yuan 2008 , takami et al . 2011 , ybarra et al . 2010 , ohlendorf et al . 2012 , lee et al . such an observational capability led to the implementation of detailed interpretative tools ( smith & rosen 2005 ; neufeld & yuan 2008 ; ybarra & lada 2009 ( yl09 ) ; takami et al . 2010 ) able to analyze irac maps of molecular flows . however , while the mere jet identification is a straightforward task , the derivation of the physical conditions from such photometric maps has to be done with some caution ( noriega - crespo et al . 2004 , neufeld & yuan 2008 ) . indeed the h@xmath0 emission lines inside the irac bands come from vibrational levels corresponding to very different excitation temperatures ( from 800 k to 4000 k ) , and contamination from pah , co and atomic lines can be significant ( takami et al . 2010 ) . in addition to allow jets identification , mid - infrared irac images have a major role in finding the embedded exciting sources that could have escaped detection at shorter wavelengths . otherwise , irac observations allow to probe the continuum emission of already identified exciting sources , but in a frequency regime where most of their energy is likely emitted . particularly suited are the irac surveys of star forming regions for which a census of the young stellar population has been already taken , since they offer an immediate correlation between the identified jets and the nature of the exciting sources . a well suited test - bed is the cloud d of the vela molecular ridge ( hereinafter vmr - d , murphy & may 1991 ) , an active site of star formation in the galactic plane showing a rich phenomenology associated with star formation processes , such as outflows ( wouterloot & brand 1999 ; elia et al . 2007 ) , jets ( lorenzetti et al . 2002b ; giannini et al . 2005 ) , and evidences of clusters as well as isolated protostars ( massi et al . 2000 , 2003 , giannini et al . 2007 , strafella et al . 2010 , hereafter sec10 ) . in the present paper we analyze the irac maps of vmr - d in order to : _ i _ ) obtain an irac - based census of the protostellar jets in vmr - d , _ ii _ ) compare the jets found in the irac images with those we have already detected in the h@xmath0 1 - 0s(1 ) at 2.12@xmath1 m and validate ( or not ) the applied irac method ; _ iii _ ) identify new exciting sources or provide the mid - ir photometry for those already known ; _ iv _ ) derive the morphology of the jets , making use also of data in other bands , and _ v _ ) correlate the properties of the jets with those of their exciting sources . the paper has the following structure : we first briefly present the data ( sect.2 ) and explain how jets are identified and analyzed ( sect.3 ) . in sect.4 we search for the jet exciting sources and present their photometry . in sect.5 we discuss our results , that are briefly summarized in sect.6 . finally , we give in appendix[sec : appendix ] a brief description of each detected jet . we have observed vmr - d as part of the irac-_spitzer _ cycle 3 gto ( pid 30335 , pi : g. fazio ) . the observations cover a region of about 1.2 square degree delimited in longitude by 264@xmath6 and in latitude by 0@xmath7 . we have presented the photometric results in sec10 , who describe in detail the observational strategy , source extraction and photometry . this work resulted in a catalog , which we use in the present paper to identify the jet exciting sources ( es ) . in the same catalog it is also listed the mips-_spitzer _ photometry at 24@xmath1 m and 70@xmath1 m ( giannini et al . 2007 , sec10 ) as well the 1.2 mm photometry ( massi et al . 2007 , referred also as simba photometry ) , which are used here to construct the sed of the exciting sources . to compare the _ maps with the h@xmath0 emission in the near - infrared , we use images of the h@xmath0 1 - 0s(1 ) ro - vibrational transition at 2.12@xmath1 m we obtained on december 2001 with sofi ( lidmann et al . 2006 ) at eso - ntt ( la silla , chile ) . these images cover the peaks of the dust continuum emission found by massi et al . ( 2007 ) and have been already searched for evidences of shocked emission by lorenzetti et al . ( 2002b ) and de luca et al . ( 2007 ) . out of 18 h@xmath0 sofi fields , we have detected shocked emission in the 11 fields displayed in figure[all : fig ] , which are overlaid on the irac image of vmr - d at 4.5@xmath1 m ( left panel ) . in the same figure , right panel , the dust emission image at 1.2 mm is depicted , as well . in one case ( jet # 15 ) the 2.12@xmath1 m image has been obtained with isaac ( cuby et al . 2004 ) at eso - vlt ( paranal , chile ) , and reported in giannini et al . ( 2005 ) . in order to supplement the photometry of the es of the jets in the wavelength ranges adjacent to those probed with irac / mips , we used the 2mass ( skrutskie et al . 2006 ) photometry in the @xmath8 , @xmath9 and @xmath10 bands , and the recently delivered data of the _ wide - field infrared survey explorer _ ( wise , wright et al . 2010 ) taken at 3.4@xmath1 m , 4.6@xmath1 m , 12@xmath1 m , and 22@xmath1 m . finally , data obtained at 250@xmath1 m , 350@xmath1 m , and 500@xmath1 m with the balloon - borne large - aperture submillimeter telescope ( blast , pascale et al . 2008 ) and compiled by netterfield et al . ( 2009 ) and olmi et al . ( 2010 ) have also been considered in the sed s building . @xmath11co(3 - 2 ) maps of four jets ( namely # 1 , 2 , 3 , 5 ) were obtained from september to december 2006 with the apex-2a heterodyne receiver , mounted on the apex telescope ( gsten et al . 2006 ) at llano chajnantor , chile . the maps are typically 40 arcseconds wide with a velocity channel width of 0.5 km s@xmath12 . the analysis of these maps will be presented in a forthcoming paper , while here they will be only marginally used in sect.[sec : jetext ] for a comparison with the irac maps . m ( relative intensity is given in the bottom bar ) . white boxes represent the fields imaged with sofi at 2.12@xmath1 m ( isaac for jet # 15 ) where jets have been discovered ( represented with red lines ) . numbered labels refer to the jet identification numbers listed in table[tab : tab1 ] . _ right panel _ - map of the dust continuum at 1.2 mm ( adapted from massi et al . 2007 ) . the sofi fields ( here shown in red ) cover most of the brightest peaks . intensity is given in the bottom bar in units of mjy / beam . [ all : fig],width=351 ] - [ 4.5 ] vs. [ 4.5 ] - [ 5.8 ] two colors plot indicating the region occupied by shocked h@xmath0 emission . constant density ( dotted ) and temperature ( solid ) lines are indicated . inside the diagram three regions are defined : _ zone 1 - _ region encompassed by curves with 10@xmath13 @xmath14h@xmath15 @xmath16 and 2300 k @xmath17 4000 k and having [ 4.5 ] - [ 5.8 ] @xmath3 0.8 mag ; _ zone 2 - _ region possibly affected by contamination of class 0/i objects : i.e. within the curves with 10@xmath2 @xmath18(h@xmath15 @xmath16 and 1500 k @xmath17 2300 k and with [ 4.5 ] - [ 5.8 ] @xmath19 0.8 ; _ zone 3 - _ region empirically determined by gutermuth et al . ( 2008 ) to contain outflows ( dashed - dotted line , i.e. roughly the portion with [ 4.5 ] - [ 5.8 ] @xmath3 0.8 mag and [ 3.6 ] - [ 4.5 ] @xmath20 1.05 mag ) . the reddening vector , corresponding to a@xmath21=30 mag , is depicted as well . [ yl_zones : fig],width=453 ] . colors ( green , red and blue ) are those corresponding to the three zones ( 1,2,3 , respectively ) depicted in figure [ yl_zones : fig ] . _ right panel _ - same portion of the sky depicted in a three - color map obtained by combining the irac images in the three bands . the blue channel is used at 3.6@xmath1 m , the green one at 4.5@xmath1 m , and the red one at 5.8 @xmath1 m . note that , while the shock - mask evidences two possible jets in orthogonal directions , the three - color image confirms the presence only of the jet in the vertical direction ( see text for more details).[ircolcol : fig ] ] yl09 have shown how the analysis of _ spitzer _ data can be used to discover and characterize the emission from protostellar outflows . such a characterization by means of the irac color - color plot relies on the presence of strong ( 0 - 0 ) and ( 1 - 0 ) h@xmath0 emission features in the irac bands ( neufeld & yuan 2008 ) , and makes it possible to define , on a [ 3.6 ] - [ 4.5 ] vs. [ 4.5 ] - [ 5.8 ] irac diagram , the region occupied by shocked h@xmath0 as a function of the gas kinetic temperature and hydrogen density , in the range 1500 k @xmath22 4000 k and 10@xmath2 @xmath16 @xmath23(h)@xmath24 10@xmath25 @xmath16 , respectively . in figure [ yl_zones : fig ] ( adapted from figure 1 of yl09 ) , we show three regions ( jet _ locus _ ) identified by yl09 to contain jets : ( 1 ) the region corresponding to jets characterized by sufficiently high temperature and hydrogen density , that are unambiguously distinguished from stellar objects ; i.e. the region encompassed by curves with 10@xmath13 @xmath14h@xmath15 @xmath16 and 2300 k @xmath17 4000 k and having [ 4.5 ] - [ 5.8 ] @xmath3 0.8 mag , zone 1 ; ( 2 ) the region possibly affected by contamination of class 0/i objects : i.e. within the curves with 10@xmath2 @xmath14h@xmath15 @xmath16 and 1500 k @xmath17 2300 k and with [ 4.5 ] - [ 5.8 ] @xmath19 0.8 mag , zone 2 ; ( 3 ) the region empirically determined by gutermuth et al . ( 2008 ) to contain outflow , i.e. roughly the portion with [ 4.5 ] - [ 5.8 ] @xmath3 0.8 mag and [ 3.6 ] - [ 4.5 ] @xmath20 1.05 mag , zone 3 . although the irac color - color plot represents a useful tool to search for jets , it has to be used with some caution , since it does not identify jets uniquely . indeed , together with the contamination of zone 2 ( see above ) , extinction effects can be significant in moving points inside and outside the jet _ locus_. we show in figure [ yl_zones : fig ] the extinction vector for a@xmath21 = 30 mag ; notably , even for extinction values lower than that , classii sources may shift inside the jet _ locus_. moreover , possible emission of the co v=1 - 0 fundamental transition at 4.6@xmath1 m and [ feii ] a@xmath26f - a@xmath27d lines between 4.4@xmath1 m and 4.9@xmath1 m can enhance the flux in band 2 , moving the colors toward the top - left corner of the diagram . therefore , as also suggested by yl09 themselves , the evaluation of the irac colors was complemented by the visual inspection of a three - color map of the region ( constructed out of the 3.6@xmath1 m , 4.5@xmath1 m , and 5.8@xmath1 m images ) to examine the morphology of the emission in the pixels falling in the jet _ locus_. in more detail , this part of the analysis was divided in two steps . first , the median subtracted irac images were converted into a color - map by applying the photometric corrections for extended sources suggested by the _ spitzer _ science center website for low surface brightness measurements ( correction factors of 0.91 , 0.94 and 0.71 at 3.6 , 4.5 and 5.8 @xmath1 m , respectively ) . as a result , we have obtained a mask for each of the three regions previously defined , which were then combined to obtain a false color representation ( called shock - mask ) depicted for all the pixels having irac colors consistent with shocked h@xmath0 . as an example , we show in figure[ircolcol : fig ] , left panel , the enlargement of the shock - mask in the region of jet # 4 , where the color code is : zone 1 : green , zone 2 : red , zone 3 : blue . as a second step , we constructed a three - color image by combining the 3.6@xmath1 m ( blue channel ) , the 4.5@xmath1 m ( green channel ) and the 5.8@xmath1 m image ( red channel ) , with the aim to confirm ( or not ) by means of a visual inspection the possible shock - excited emission localized in the shock - mask . first , we discard all pixels that , although falling in the shock - mask , appear in the three - color image as isolated pixels or structures composed by less than 4 - 5 pixels which are likely due to artifacts or noise fluctuations . then , we applied the following prescriptions to confirm groups of pixels in the shock - mask as real knots : * they must appear as extended structures . in sec10 , we define as point - like those sources with a shape parameter @xmath28 , where @xmath29=@xmath30(observed ) - @xmath31(psf ) , is the difference between the squares of the half - width of the best - fitting function and the point - spread function . here , we request that knots must have @xmath32 0.7 . this prescription , although implying that possible point - like knots are discarded , has however the advantage of easily individuate protostellar sources in the jet _ locus _ of the color - color diagram . * preferably , knots should belong to a chain ( which can have a straight- , curved- , or @xmath33-shape ) . they must be detected at signal - to - noise level greater than 3 in at least one band . * isolated knots are considered only if detected at signal - to - noise level greater than 3 in at least two bands . in figure[ircolcol : fig ] , right panel , we show the enlargement of the three - color image in the region of jet # 4 . this is an illustrative case of the importance of the visual inspection described above : while the shock - mask in the left panel evidences two possible jets in orthogonal directions , the visual inspection of the three - color image confirms the presence only of the jet in the vertical direction , since the pixels in the horizontal direction correspond to image artifacts , to point - like sources , or to pixels whose photometry is strongly contaminated by the emission of the central cluster . m : blue - 4.5@xmath1 m : green - 5.8@xmath1 m : red ) irac image . h@xmath0 knots locations are highlighted with boxes and labeled as in table[tab : tab2 ] . the candidate exciting source(s ) is indicated with a circle . _ right panel _ - three colors ( 3.6@xmath1 m : blue - 4.5@xmath1 m : green - 24@xmath1 m : red ) irac / mips image where the contours of the dust emission at 1.2 mm ( massi et al . 2007 ) are superposed in steps of 4@xmath34 starting from a level of 50 mjy / beam . the candidate exciting source(s ) is ( are ) indicated with a circle.[jet_1:fig],width=680 ] for jet 2.[jet_2:fig],width=680 ] for jet 3.[jet_3:fig],width=680 ] for jet 4.[jet_4:fig],width=680 ] for jet 5.[jet_5:fig],width=680 ] for jet 6.[jet_6:fig],width=680 ] for jet 7.[jet_7:fig],width=680 ] for jet 8.[jet_8:fig],width=680 ] for jet 9.[jet_9:fig],width=680 ] for jet 10.[jet_10:fig],width=680 ] for jet 11.[jet_11:fig],width=680 ] for jet 12.[jet_12:fig],width=680 ] for jet 13.[jet_13:fig],width=680 ] for jet 14.[jet_14:fig],width=680 ] for jet 15.[jet_15:fig],width=680 ] with the procedure described in the previous section , we have found a total of 15 jets inside the area observed with irac , as well as some sparse knots that do not show any clear alignment . noticeably , we re - confirm with irac all the jets that were already detected in the near - infrared , except jets # 7 and # 14 that represent new discoveries . this is depicted in figure[all : fig ] , where it is clear that almost all the _ spitzer _ jets ( red lines ) are located within a field observed in the h@xmath0 2.12@xmath1 m line . indeed , such result is not unexpected , since h@xmath0 _ spitzer _ and 2.12@xmath1 m lines , coming from levels with similar excitation energy , probe similar excitation conditions . conversely , some knots ( within the jets ) detected in the 1 - 0 s(1 ) line with sofi do not appear in any _ spitzer _ image . in the case of jet # 15 , this occurs because the most internal knots ( e.g. knots d and f reported by giannini et al . 2005 ) are confused in the _ spitzer _ images in the strong diffuse emission of the protostellar cluster where the exciting source is located . in other cases of faint knots ( at 2.12@xmath1 m ) , the lack of a _ spitzer _ counterpart can be ascribed to the poorer sensitivity at longer wavelengths . for example , the 3@xmath34 sensitivity limit at 2.12@xmath1 m of @xmath35 0.01 mjy is around a factor of 3 better than the sensitivities at 3.6@xmath1 m and 4.5@xmath1 m . the very nice coincidence between irac and sofi jet detections leads to three main conclusions : _ i _ ) the approach adopted to discover jets in the irac images can be considered as validated , _ ii _ ) h@xmath0 2.12@xmath1 m surveys allow to catch a relevant fraction of jets and hence longer wavelengths observations are not mandatory to overcome extinction problems , and , _ iii _ ) the absence of jets in the parts of the irac map not covered also in the near - infrared strongly indicates that jets can be preferentially found in proximity of the dust peaks , namely where the near - infrared images were originally taken ( see also de luca et al . 2007 ) . the 15 _ spitzer _ jets are depicted in figures[jet_1:fig ] to [ jet_15:fig ] and are described in appendix [ sec : appendix ] . in each of these figures , we show in the left panel a composite irac 3-color image and in the right panel a composite irac - mips image with overlaid the contours at 1.2 mm ( these latter in steps of 4@xmath34 starting from a 4@xmath34 level of 50 mjy / beam ) . in these figures , jet knots are marked by small boxes while the candidate exciting source , determined as described in sect.[sec : sec4 ] , is marked by a small circle . a summary of the jet properties is given in table [ tab : tab1 ] , where for each jet we give in columns 1 - 4 the identification number , the number of the detected knots , the projected length , and the position angle . as anticipated , all the jets were originally discovered in the h@xmath0 2.12@xmath1 m line , the only exceptions being represented by jet#7 , # 14 and by other few knots ( k1-k4 ) , which were not targeted in the near - infrared . in the last column , we give the references where these jets have been reported . the projected jet lengths and position angles are in agreement with those given in the literature , the only exception being the length of jet#6 , which was originally supposed to have a double length with respect to that reported in table[tab : tab1 ] ( giannini et al . this hypothesis was done since the jet appears one - sided in the 2.12@xmath1 m image , and the lack of a counter - jet was ascribed to the limited field of view of sofi . based on the irac image ( see fig.[jet_6:fig ] ) we now confirm the lack of any additional knot symmetric to the candidate es . as a general note , jets # 2,3,7 , and 11 are rather compact ( length lower than 0.15 pc ) , jets # 4,14 , and 15 extend for more than 0.4 - 0.5 pc , while all the others have intermediate lengths . length of jets # 5 and # 13 is very uncertain , since it is not clear whether or not the discovered knots can be effectively attributable to the same jet ( see also the discussion in appendix[sec : appendix ] ) . regarding the position angle ( pa , computed from the north to the east ) , notably around half of the jets are oriented 160@xmath36 : we will comment this feature in sect.[sec : sec5.4 ] . to obtain the photometry of the detected knots we have applied the task _ polyphot _ within the image reduction and analysis facility ( iraf ) package , which is suited for evaluating the emission of extended and irregular sources . for each knot ( both in the irac bands and in the narrow - band at 2.12 @xmath1 m ) , we have first defined a polygon with sides tangent to the 3@xmath34 contour of the emission and then summed up the flux of all the pixels inside it . the sky background was estimated locally and then subtracted from the flux computed inside the polygon . as shown by neufeld & yuan ( 2008 ) , the photometry in the irac bands are dominated by bright h@xmath0 lines , and in particular by the @xmath4=1 - 0 o(5)-o(7 ) in band 1 ( 3.6@xmath1 m ) , @xmath4=0 - 0 s(9 ) in band 2 ( 4.5@xmath1 m ) , @xmath4=0 - 0 s(7)-s(6 ) in band 3 ( 5.8@xmath1 m ) , and @xmath4=0 - 0 s(5)-s(4 ) in band 4 ( 8.0@xmath1 m ) . minor contributions are expected from pah , co and atomic lines which , however , can not be easily estimated ( takami et al . 2010 ) and thus are not taken into account by us in the photometry . similarly , we ascribe the near - infrared narrow - band filter photometry totally to h@xmath0 1 - 0 s(1 ) line emission , since no continuum emission in the @xmath37-band is usually seen in the spectra of protostellar jets ( e.g. giannini et al . 2004 , nisini et al . table[tab : tab2 ] lists the coordinates and the size of each knot , evaluated on the 4.5@xmath1 m image , along with the photometry between 2.12@xmath1 m and 8.0@xmath1 m . the last two columns list the irac luminosity ( @xmath38 , obtained by summing up the luminosities of all the irac bands ) and 2.12@xmath1 m luminosity ( @xmath39 ) , both computed assuming a distance to vmr - d of 700 pc ( liseau et al . as a second goal of our study , we have used the irac and mips images to search for the es using the following procedure . * on the irac image at 4.5@xmath1 m we have defined a rectangle oriented as the jet itself and having the long side equal to @xmath40 @xmath41 the jet length , where the value of @xmath40 , in the range 1 - 3 , is selected for each jet to not exceed the parsec scale . the short side , orthogonal to the jet direction , is defined case - by - case so that all the knots lie inside the rectangle . * we then searched the irac catalog ( sec10 ) for all the point - like sources inside the rectangle , selecting those whose position is compatible with the jet shape ( straight- curved- or _ s_-shape ) . * sources are then selected if they have a spectral index ( computed by using the available measurements between @xmath10 and 8@xmath1 m ) compatible ( within the upper limits ) with a yso of classii , i or flat . to class i or flat sources we give higher priority than to nearby class ii sources . analogously , we have searched the mips catalog to select possible sources not present in the irac catalog because not detected at wavelengths shorter than 24@xmath1 m . the results of this search are presented in table[tab : tab3 ] . we remark that , given the crowding of ysos in the investigated fields , we can not consider our search exhaustive , especially for the presence of many classii sources in the defined rectangles . consequently , the sources listed in table[tab : tab3 ] have to be considered as es _ candidates _ and hence as interesting objects to be further studied . noticeably , however , we find only one best candidate in all cases , apart from jets # 5 and # 14 , for which two possible candidates have been selected ( labeled with _ a _ and _ b _ according with a priority order , see appendix[sec : appendix ] ) . associations with counterparts at different wavelengths ( from 1.2@xmath1 m to 1.2 mm ) are reported , with the corresponding identification number assigned in each catalog . associations between irac , mips and simba sources have been already done by sec10 and reported in their photometric catalog ; here we also consider possible associations with 2mass , wise and blast sources . these have been matched to the irac source list by using a circle whose radius is the sum of the positional uncertainties associated with each pair of instruments ( reported in the caption of table[tab : tab3 ] ) . the es of jets # 2 , # 3 , # 11 , # 13 are very barely or not detected at wavelengths shorter than 24@xmath1 m ; this confirms the results presented in giannini et al . 2007 ( for jets # 2 , # 3 ) who hypothesized them as very young ysos . all except one sources have a mips 24@xmath1 m counterpart and 8 ( 10 ) are associated with a 250@xmath1 m ( 1.2 mm ) peak , while only 5 objects are seen in the 2mass bands . the source exciting jet#15 has been already the subject of a dedicated study ( giannini et al . it is a member of an infrared cluster that appears completely unresolved and saturated in the _ spitzer _ bands . for this reason that object ( also known as irs17- # 40 - 3 ) will not be commented further on in the present paper . in table[tab : tab4 ] we list the fluxes ( in mjy ) of the es , which will be used in sect.[sec : seds ] to derive their sed . as far as the wise data are concerned , we give only the fluxes at 12@xmath1 m ( band 3 ) . those in bands 1,2,4 ( at 3.4 @xmath1 m , 4.6 @xmath1 m , and 22 @xmath1 m ) do not differ from the _ spitzer _ ones ( at 3.6@xmath1 m , 4.5@xmath1 m , and 24@xmath1 m ) by more than 30% , which does not imply significant modifications in the sed global shape . two sources ( # 4 and # 14a ) are saturated at 24@xmath1 m , hence the wise4 band flux has been used for their sed fitting . finally , in the last column of table[tab : tab4 ] we list the source spectral index , computed for sources with at least four valid fluxes between 2.16@xmath1 m and 12@xmath1 m . as a further step toward identifying the driving sources , we have investigated whether the selected candidate es of each jet is located at the center of two distinct lobes traveling in opposite directions , as expected for bipolar jets . since we lack high - resolution spectroscopy observations , which would have allowed us to directly measure the mutual shift of the emission lines , we have adopted an indirect way to identify ( if any ) the blue- and red - shifted lobes . this method is based on the comparison of the irac total luminosity ( @xmath38 ) with the luminosity of the 1 - 0 s(1 ) line ( @xmath39 ) , both listed in the last two columns of table[tab : tab2 ] . such comparison helps in identifying the blue- and red - shifted lobes under some conditions : _ i ) _ the visual extinction increases in the direction of the red lobe ( as expected if this coincides with the receding part of the jet and if this latter is not in the sky - plane ) ; _ ii ) _ @xmath39/@xmath38 weakly depends on the temperature and density in a given knot . + to verify the condition _ ii ) _ , we have estimated the ratio @xmath39/@xmath38 for temperatures between 1000 k and 3000 k , in the local thermal equilibrium ( lte ) approximation . as shown in figure[lte : fig ] , such ratio remains fairly constant in the range @xmath35 0.2 - 0.4 ( note that this is generally not true if the ratio of @xmath39 with the luminosity in a single irac band is taken ) . hence , a difference of @xmath39/@xmath38 of more than a factor of @xmath35 2 in symmetrically located knots ( with respect to the putative exciting source ) can be primarily ascribed to a difference in the local extinction ( @xmath42a@xmath21 ) , since this latter affects the near - infrared and the mid - infrared fluxes in a differential way . ( we note , however , that the assumption _ ii ) _ is too coarse to allow the absolute determination of a@xmath21 ) . + in table[tab : tab5 ] we list for each jet the pairs of symmetric ( groups of ) knots for which we have evaluated @xmath42a@xmath21 , computed by assuming the rieke & lebofsky ( 1985 ) extinction law . to minimize the uncertainties , this has been obtained from the average @xmath39/@xmath38 in each side of the jet . adding the photometric uncertainties to the allowed variations in @xmath39/@xmath38 due to the physical conditions in the knots , we consider as significant only @xmath42a@xmath21 larger than 2 mag , which roughly correspond to a difference of @xmath39/@xmath38 of more than 4 . summarizing the results , our objects are divided in three groups , in which we measure a @xmath42a@xmath21 with a decreasing level of significance . in particular we get : @xmath42a@xmath21 @xmath435 mag in five jets ( # 1 , 2 , 4 , 8 , 11 ) , 2 mag @xmath44 @xmath42a@xmath21 @xmath24 5 mag in two jets ( # 3 , 12 ) and @xmath42a@xmath21 @xmath24 2 mag in four jets ( # 5 , 9 , 10 , 15 ) . for jet # 13 we measure @xmath42a@xmath21 both considering knots 3 and 4 ( @xmath42a@xmath21= 10 mag ) and the groups of knots 1 - 2 and 5 - 6 ( @xmath42a@xmath21= 4 mag ) . in both cases all the knots south ( north ) the es are red- and blue - shifted , respectively . this result slightly favors the hypothesis that all the knots are part of the same jet ( see also appendix[sec : appendix ] ) . for the remaining jets we are not able to apply the method either because we can not identify knots symmetrically located ( jet # 6 ) or because measurements of the 2.12@xmath1 m line do not exist ( jets # 7 , 14 ) . in particular , for jet # 14 ( as well as for jet # 5 , where @xmath42a@xmath21 is not significant ) we can not use the extinction method to identify the exciting source between the two candidates labeled with @xmath45 and @xmath46 in table[tab : tab3 ] . remarkably , the few direct velocity measurements existing in the literature for the vmr - d jets are all in agreement with the results obtained with the above extinction method. these are the high - resolution spectroscopy observations of the 2.12@xmath1 m line of jet # 15 ( giannini et al . 2005 ) , the @xmath11co(1 - 0 ) , @xmath47co(2 - 1 ) maps in the vicinity of the iras sources and overlapping jets # 4 , 8 , and 15 ( elia et al . 2007 ) , and the small @xmath11co(3 - 2 ) maps obtained with apex for jets # 1 , 2 , and 3 ( see as an example the map of jet # 3 shown in figure[fig : apex ] ) . another @xmath11co(3 - 2 ) map ( although incomplete ) of jet # 5 should favor source 5b as the exciting source , since all the parts of the outflow overlapping knots 2 - 7 appear blue - shifted . once the observed fluxes have been determined to construct the seds for our candidate es , we used the radiative transfer model developed by whitney et al . 2003a , whitney et al . 2003b to obtain insight on the physical properties of these objects . actually , the use of this model has been greatly facilitated by robitaille et al . 2006 , robitaille et al . 2007 , who provided a grid of 200,000 computed models and an efficient online fitting procedure to help the community in the sed fitting work . in the best - fit procedure we set as constraints the extinction between 0 mag and 200 mag and the distance to the vmr - d within the range 500 pc - 1000 pc ( liseau et al . 1992 ) . we fit the sed of sources with at least six photometric detections in table[tab : tab4 ] . in addition , a single grey - body function was used to fit the sed of the sources not observed at wavelengths shorter than 24@xmath1 m , but with at least 3 data points ( # 2 , # 3 , # 13 ) . we used different - sized apertures with radius corresponding to 3@xmath34 of the instrument beam and verified that even adopting larger apertures the same models are selected . the results of the best - fits a la robitaille are depicted in figure[seds : fig ] . here we show as upper limits also the blast ( 350 - 500@xmath1 m ) data points , although not considered in the fitting procedure . this choice is motivated by the fact that multiple sources lie in the large blast beam . the main output parameters are reported in table[tab : tab6 ] , where we give both the values of the best - fit model ( in boldface ) and those of the model whose @xmath48 is 50% higher than the minimum value . these parameters are the distance to the source ( col.2 ) , the disk inclination ( between 0@xmath49 and 90@xmath49 with respect to the sky - plane , col.3 ) , the interstellar and circumstellar extinction ( cols.4,5 ) , the stellar temperature and mass ( cols.6,7 ) , the outer envelope radius ( col.8 ) along with its mass ( col.9 ) , and the total ( stellar plus accretion ) luminosity ( col.10 ) . the general picture emerging from the seds modeling is that most of the es are low - mass and relatively evolved protostars . the stellar temperatures ( 2900 k @xmath50 4400 k ) and masses ( typically @xmath51 0.05 - 1 m@xmath52 ) are those typical of embedded ( a@xmath21 @xmath35 10 - 60 mag ) t tauri stars of spectral type between k0 and m0 . the reservoir of material of the surrounding envelope ( @xmath53 typically between 10@xmath54- a few m@xmath52 ) is not enough to enhance significantly the final stellar mass . possible exceptions to this general picture are : _ i ) _ sources # 1 and # 4 , whose mass ( already piled up onto the star or still in the envelope ) appears potentially sufficient to form an intermediate - mass object . given the crowding of the regions where these sources are located ( close to infrared clusters , see also table[tab : tab1 ] ) , we can not exclude however that they are binaries or multiple objects unresolved at the irac angular resolution ; _ ii ) _ sources # 7 and # 14a , for which the fitted @xmath53 and @xmath55 are very uncertain , since their sed is not measured beyond @xmath56=70@xmath1 m . finally , we note that the disk inclinations of sources # 1 , # 4 , # 8 , for which the correspondent jets have two mutually shifted lobes ( i.e. for which @xmath42a@xmath21 is @xmath43 5 mag , see sect.[sec : jetext ] ) , are 20@xmath49 - 40@xmath49 , 55@xmath49 - 65@xmath49 , 20@xmath49 . in the hypothesis that jet and disk are orthogonal , these inclinations confirm that these jets lie outside the sky - plane . in figure[fig : jetcolor ] the projected jet length ( in pc ) is displayed as a function of the far - ir color [ 24]-[70]-[70 ] color is defined as 2.5log(@xmath57 ) where @xmath58,@xmath59 are the fluxes at 24 , 70 @xmath1 m , respectively and @xmath60,@xmath61 the zero magnitude fluxes . ] of the candidate es . this plot is based only on sources selected as first priority es ( see also appendix[sec : appendix ] ) . the idea underlying such plot is to investigate whether or not a relationship exists between the two considered parameters within an evolutionary framework . indeed , the youngest jets are expected to be the shortest , in the assumptions that all the jets travel at comparable velocities and that their inclinations with respect to the sky - plane are not dramatically different . in this sense the following analysis has to be considered in a statistical sense , and no conclusions can be drawn for the individual sources . the [ 24]-[70 ] color is typically used as an age indicator , because the youngest sources are also the coldest and reddest ones . figure [ fig : jetcolor ] shows an evident correlation ( regression coefficient @xmath43 0.8 ) which well agrees with the expectations . noticeably , the same kind of plot constructed with irac ( instead of mips ) colors or with the spectral index between 2@xmath1 m and 12@xmath1 m does not show any correlation : bluer colors are , in fact , more sensitive both to different circumstellar morphologies ( e.g. disk inclination ) and to extinction effects . on the other hand , the [ 24]-[70 ] color is more sensitive to the circumstellar envelope temperature and almost unaffected by extinction , since it changes of only 0.2 mag for a@xmath21 @xmath62 100 mag . hence , the observed correlation is compatible with an evolutionary interpretation , and can not be related to a mere inclination effect : if this latter were responsible for the shortening of the jet projected length , the bluest ( warmest ) portions of the circumstellar disk material would be exposed to the observer , hence an anti - correlation would emerge between the jet projected length and the [ 24]-[70 ] color . in figure[fig : lumlum ] we plot the irac line cooling as a function of the es bolometric luminosity ( considering only es of higher priority ) . our aim is to investigate the balance between the energy irradiated by the driving protostar and that emitted by the associated jet . in that figure , the horizontal dotted lines are the range of bolometric luminosity given in table[tab : tab6 ] , with the value associated with the best - fit model indicated with a dot . the reported values of @xmath38 are those given in table[tab : tab2 ] without any extinction correction ; for reference , these values increase by a factor of 0.3 and 1.3 ( in logaritmic scale ) for a@xmath21 = 10 mag and 30 mag , respectively . instead , the uncertainties on @xmath38 associated with the photometric errors are not significant . a rough linear correlation ( in the log - log scale ) is found between the two quantities ( regression coefficient of 0.68 ) , with an intercept value located at @xmath38 @xmath35 10@xmath63 l@xmath64 ) . this value can be easily converted in the total h@xmath0 cooling ( @xmath65 ) by estimating the ratio @xmath66 under lte conditions : we find that this ratio is @xmath35 4 at temperatures of 2000 - 3000 kelvin . hence , the above relation translates into @xmath65 @xmath35 10@xmath67 l@xmath64 . this value is in good agreement with the result ( @xmath65 @xmath35 10@xmath68 l@xmath64 ) found by caratti o garatti et al . ( 2006 ) in a sample of class0/i sources . in that work no significant differences in terms of efficiency were found between the two classes , likely because of the large uncertainties associated with @xmath65 , which was largely derived on the base of near - infrared observations . these differences are indeed expected in the commonly accepted scenario in which a tight relationship exists between mass - loss and mass accretion rate , with a declining of the latter with the source evolution ( e.g. bontemps et al . 1996 ) , in figure[fig : lumlum ] a certain degree of segregation between the points above ( sources # 1 , 3 , 4 , 6 , 9 , and 13 ) and below the straight line ( sources # 2 , 7 , 8 , 10 , and 14a ) is indeed recognizable , the formal fits through the two groups having regression coefficients of 0.97 and 0.99 , respectively . we estimate that the two groups can mix each other only if their differential extinction is ( on average ) larger than 15 - 20 mag . even a more clear segregation is found by using far - infrared lines . for example , the ratio between @xmath69 ( i.e. the cooling rate due to the far - infrared lines ) with @xmath70 gives values spanning from 10@xmath54 to 10@xmath71 going from class0 to classii sources ( nisini et al . 2002 , lorenzetti et al . 2002a ) . the circumstellar accretion disks are expected to have their symmetry axis parallel to the local magnetic field ( shu et al . 1995 ) . at larger scales , the magnetic field is expected to have a role only in regions located in the galactic plane and far from the center , i.e. where it is oriented parallel to the galactic disk ( 150@xmath72 - 160@xmath72 , stephens et al . indeed , jets with orientations parallel to the large - scale magnetic field have been already found both _ in _ the galactic plane ( hodapp 1984 , 1987 ; scarrott et al . 1992 ; lorenzetti et al . 2002b ) and _ outside _ of it ( in taurus - mnard & duchne 2004 ) . given its location , vmr - d represents a well suited case to test the influence of the galactic magnetic field on the jet orientation . remarkably , out of 15 jets , 8 ( # 1 , 4 , 6 , 8 , 9 , 11 , 13 , 15 ) have a position angle of @xmath35 160@xmath73 15@xmath72 ( see table[tab : tab1 ] ) . for completeness , also the jet from irs8 ( located in vmr , as well ) presents an orientation of @xmath35 153@xmath72 ( lorenzetti et al . hence , the present survey suggests that the galactic magnetic field may be important during the early phases of the pre - main sequence evolution . however , a selection effect could bias our sample and make this inference weaker : since the influence of the galactic magnetic field tends to decrease toward the inside of the cloud , we expect that jets aligned with it are located preferentially in the external layers . therefore , these jets are likely the least embedded and therefore those most easily detected . the present h@xmath0-jet survey covers a rather small region ( about 1.2 deg@xmath2 ) , but it has the advantage of being substantially unbiased and therefore complete at the _ spitzer _ sensitivity . hence it is worthwhile to compare our results with those stemming from other similar surveys that , having both investigated larger areas and detected a larger number of sources , offer a more robust statistical significance in tracing the early phases of the stellar formation . indeed , the peculiarity of our survey is to study a region inside a gmc , which is the typical site of the star formation , but rather different from other nearer dark clouds where star formation occurs in a usually less turbulent environment . there are currently two main projects regarding 2.12@xmath1 m h@xmath0 jet emission : _ ( i ) _ the survey uwish2 ( ukirt widefield infrared survey of h@xmath0 , froebrich et al . 2011 ) ; and _ ( ii ) _ mhos ( molecular hydrogen emission - line objects ) , namely the general catalog of h@xmath0 infrared ( 1 - 2.2 @xmath1 m ) emission from young outflowing objects ( davis et al . 2010 ) . on both papers references are given to the numerous 2.12@xmath1 m h@xmath0 imaging surveys of different regions carried out in the past years . since uwish2 will eventually provide an unbiased and complete census of the 1 - 0s(1 ) ro - vibrational h@xmath0 line emission along the galactic plane , beside star forming and hii regions , it will include also evolved ( agb ) stars , supernova remnants and fluorescent h@xmath0 emission . conversely , mhos catalog is a fundamental database only of ysos jets and hh objects that present molecular emission . some of the jets described in the present paper have been already mentioned in a very preliminary form ( giannini et al . 2007 ) , and , as such , are included in mhos . finally , an example rather similar to our vmr - d survey ( as far as its gmc nature , size of the investigated area , distance from the sun are concerned ) is represented by the extensive orion a survey ( stanke et al . some results from the literature are the following : * \(a ) a potential source candidate can be assigned to about half of the jets ; ( b ) typically , the flows are clustered in group of 3 - 5 objects , within a radius of 5 pc ; ( c ) these groups are separated by about 5 pc ; ( d ) 2/3 of the surveyed area is devoid of jets ; ( e ) a large range of apparent lengths is found ( from 4 to 130 arcsec ) ; ( f ) a small fraction ( about 10% ) are `` parsec scale '' jet ; ( g ) there is a marginal ( less than 3@xmath34 ) over - abundance of flow position angles roughly perpendicular to the galactic plane . all of these are preliminary considerations based on the serpens and aquila survey ( ioannidis & froebrich , 2012 ) . * \(a ) in some regions multiple knots and bow shocks radiate in many directions from a tight cluster of young stars and the relationship between these objects is often unclear , mainly in massive star forming clusters ; ( b ) jets are found in both low- and high - mass star forming regions . * \(a ) the first outcome of this survey is the large number of detected jets distributed in a more isolated way with respect to orion b that presents a more clustered distribution ; ( b ) a large variety of morphologies have been detected , ranging from large , extended , filamentary features to compact or even unresolved knots . this circumstance makes the identification of flows not always an objective process ; ( c ) for a reasonable fraction of the jets a candidate exciting source is either obvious or suggested ; ( d ) few jets show some degree of symmetry ; ( e ) very few jets are well collimated , virtually unresolved orthogonally to the flow direction ; ( f ) apparently , there is not a preferred flow orientation . the results of our vmr - d survey differ in several respects from the previous surveys findings , mainly because these are based only on near - ir data . in the majority of cases ( 13 out of 15 ) a convincing exciting source has been found : this circumstance is related to the increased capability offered by the longer wavelengths coverage to identify young and embedded objects . practically , all the vmr - d jets display a high degree of collimation and most of them are composed of knots arranged in symmetrical position . preserving the collimation and a certain symmetry over a large part of the gmc is a key - feature that makes easier the identification of a candidate driving source and it could signal the presence of a relatively quiet environment , not too much affected by ( local ) turbulence . in this sense , vmr - d could be different from both the more active vmr - a ,- b ,- c clouds and orion gmc ; indeed , this picture is also supported by independent evidences as the lack in vmr - d of o - b associations ( liseau et al . 1992 ) and the overall gas dynamics ( olmi et al . 2010 ) . this paper presents a systematic search of molecular jets from young protostars in the star formation region vela , cloud d. our color - color analysis in the four bands of _ spitzer_-irac has led to the following results : * 15 jets have been detected in an area of about 1.2 square degrees , as well as a few sparse knots not displaying a clear alignment . all except two jets were previously known since already imaged in the h@xmath0 2.12@xmath1 m line . the absence of jets in the part of the irac map not covered also in the near - infrared strongly indicates that jets can be preferentially found in proximity of the dust peaks , since these are the locations where the 2.12@xmath1 m images have been taken . * using the available catalog of ysos in vmr - d we searched the jet exciting sources . for each jet , we selected all nearby classi , flat , and classii sources , and selected those at the jet center and aligned with it . in this way we are able to find a best - candidate exciting source in all except two jets , for which we remain with two possible alternatives . to validate the jet - es association , we have investigated whether a differential extinction exists between the two ( hypothetical ) lobes . through an analysis of @xmath66 , we find a significant extinction gradient of @xmath42a@xmath74 5 mag in five cases . * a search for the es counterparts at wavelengths between 1.2@xmath1 m and 1.2 mm has been done . all except one source have a 24@xmath1 m counterpart , and 10 are associated with a 1.2 mm peak . four sources are very barely or not observed at wavelengths shorter than 24@xmath1 m , suggesting they are very young protostars ; three of them are also associated with the most compact jets ( @xmath75 0.1 pc ) . * the seds of the es have been constructed based on all available photometric data between 1.2@xmath1 m and 1.2 mm . we used the robitaille et al . ( 2006 , 2007 ) model to derive the main source parameters , such as stellar temperature and mass , envelope mass and radius , and total bolometric luminosity . on average , the es are low - mass and relatively evolved protostars , with the possible exception of few objects located close to infrared clusters , which could be intermediate - mass protostars as well as binaries or multiple objects not resolved at the _ spitzer _ angular resolution . * a significant correlation is found between the projected jet length and the [ 24 ] - [ 70 ] color , which is consistent with an evolutionary scenario according to which shorter jets are associated with younger sources . we remark , however , that this trend has to be considered just in a statistical sense , and that no conclusions can be drawn for the individual sources , given the strong assumptions ( e.g. jet inclination ) under which the correlation is derived . the same trend is not recognizable if the diagram is drawn with shorter wavelengths colors , whose values are significantly affected by geometrical and extinction effects . * a coarse correlation is found between irac line cooling and the es bolometric luminosity , which is in agreement with previous literature determinations . it is suggested that the sources segregate in two distinct groups , to which pertain a different efficiency in the line cooling . this trend is in agreement with the the scenario in which mass loss and mass accretion are tightly related phenomena , with mass accretion rates decreasing with time . * about half of the jets present a position angle which may indicate a certain influence by the galactic magnetic field , at variance with other star forming regions not located exactly in the galactic plane . however , a selection bias could affect our sample . indeed , since the influence of the galactic magnetic field tends to decrease toward the inside of the cloud , we expect that jets aligned with it are located preferentially in the external layers , where they are most easily detected . * unlike other similar studies conducted on other star forming regions , the association between jet and es has been more successful in vmr - d . together with the increased capabilities offered by the longer wavelengths at which vmr - d is observed , this circumstance is likely favored by the high degree of collimation displayed by the jets . this , in turn , reflects the presence of a relatively quiet environment , for example in comparison with the other vela clouds where star formation is going on . this paper is dedicated to the memory of alberto salama who prematurely passed away during the development of the work . it is based on observations made with the _ spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa . black , j.h . , & van dishoeck , e.f . 1987 , apj , 322 , 412 bontemps . s. , andr , p. , terebey , s. , & cabrit , s. 1996 , a&a , 311 , 858 caratti o garatti , a. , giannini , t. , nisini , b. , & lorenzetti , d. 2006 , a&a , 449 , 1077 caswell , j. l. , & haynes , r. f. 1987 , a&a , 171 , 261 cuby , j. c. , lidman , c. , johnson , r. , jaunsen , a. , moutou , c. 2004 , isaac user manual cyganowski , c. j. , brogan , c. l. , hunter , t. r. , churchwell , e. , zhang , q. 2011 , apj , 729 , 124 davis , c.j . , froebrich , d. , stanke , t. , et al . 2009 , a&a , 496 , 153 davis , c.j . , gell , r. , khanzadyan , t. , smith , m.d . , & jenness , t. 2010 , a&a , 511 , a24 davis , c.j . , scholtz , p. , d. , lucas , p. , smith , m.d . & adamson , a. 2008 , mnras , 387 , 954 de luca , m. , giannini , t. , lorenzetti , d. , et al . 2007 , a&a , 474 , 863 elia , d. , massi , f. , strafella , f. , et al . 2007 , apj , 655 , 316 fazio , g.g . , hora , j.l . , allen , l.e . 2004 , apjs , 154 , 10 froebrich , d. , davis , c. j. , ioannidis , g. , et al . 2011 , mnras , 413 , 480 giannini , t. , mccoey , c. , caratti o garatti , a. , et al . 2004 , a&a , 419 , 999 giannini , t. , lorenzetti , d. , de luca , m. , et al . 2007 , apj , 671 , 470 giannini , t. , massi , f. , podio , l. , et al . 2005 , a&a , 433 , 941 gsten , r. , nyman , l. . , schilke , p. , et al . 2006 , a&a , 454 , 13 gutermuth , r.a . , myers , p.c . , megeath , s.t . 2008 , apj , 674 , 336 hodapp , k .- w . 1984 , a&a , 141 , 255 hodapp , k .- w . 1987 , apj , 319 , 842 hollenbach , d. , & mckee , c.f . 1989 , apj , 342 , 306 kaufman , m.j . , & neufeld , d.a . 1996 , apj , 456 , 611 ioannidis , g. , & froebrich , d. 2012 , mnras , 421 , 3257 lee , h .- t . , takami , m. , duan h .- y . , et al . 2012 , apjs , 200 , 2 lidmann , c. , cuby , j .- g : , vanzi , l. , et al . 2006 , sofi user s manual liseau , r. , lorenzetti , d. , nisini , b. spinoglio , l. & moneti , a. 1992 a&a , 265 , 577 lorenzetti , d. , giannini , t. , nisini , b. , et al . 2002a , a&a , 395 , 637 lorenzetti , d. , giannini , t. , vitali , f. , massi , f. & nisini b. 2002b , apj , 564 , 839 massi , f. , de luca , m. , elia , d. , et al . 2007 , a&a , 466 , 1013 massi , f. , giannini , t. , lorenzetti , d. , et al . 1999 , a&as , 136 , 471 massi , f. , lorenzetti , d. , & giannini , t. 2003 , a&a , 399 , 147 massi , f. , lorenzetti , d. , giannini , t. , & vitali , f. 2000 , a&a , 353 , 598 mnard , f. , & duchne , g. 2004 , ap&ss , 292 , 419 murphy , d.c . & may , j. 1991 , a&a , 247 , 202 netterfield , c. b. ade , p. a. r. , bock , j. j. , et al . 2009 , apj , 707 , 1824 neufeld , d.a . , & yuan , y. 2008 , apj , 678 , 974 nisini , b. , caratti o garatti , a. , giannini , t. , et al . 2002 , a&a , 393 , 1035 nisini , b. , giannini , t. , & lorenzetti , d. 2002 , apj , 574 , 246 noriega - crespo , a. , morris , p. , marleau , f. r. , et al . 2004 , apjs , 154 , 352 ohlendorf , h. , preibisch , t. , gaczkowski , b. , et al . 2012 , a&a , 540 , 81 olmi , l. , angls - alczar , d. , de luca , m. , et al . 2010 , apj , 723 , 1065 pascale , e. , ade , p.a.r . , bock , j.j . , et al . 2008 , apj , 1081 , 400 rieke , g. h. , & lebofsky , m. j. 1985 , apj , 288 , 618 robitaille , t.p . , withney , b. , indebetouw , r. , wood , k. , & denzmore , p. 2006 , apjs , 167 , 256 robitaille , t.p . , withney , b. , indebetouw , r. , & wood , k. 2007 , apjs , 169 , 328 scarrott , s.m . , draper , p.w . , rolph , c.d . , & wolstencroft , r.d . 1992 , mnras , 257 , 485 shu , f.h . , najita , j. , ostriker , e.c . , & shang , h. 1995 , apj , 455 , l155 smith , m.d . 1995 , a&a , 296 , 789s smith , m.d . & rosen , a. 2005 , mnras , 357 , 1370 smith , h.a . , hora , j.l . , marengo , m. , & pipher , j.l . 2006 , apj , 645 , 1264 stanke , t. , mccaughrean , m.j . , & zinnecker , h.d . 1998 , a&a , 332 , 307 stanke , t. , mccaughrean , m.j . , & zinnecker , h. 2002 , a&a , 392 , 23 stephens , i.w . , looney , l.w . , dowell , c.d . , vaillancourt , j.e . , & tassis , k. 2011 , apj , 728 , 99 strafella , f. , elia , d. , campeggio , l. , et al . 2010 , apj , 719 , 9 ( sec10 ) skrutskie , m.f . , cutri , r.m . , stiening , r. , et al . 2006 , aj , 131 , 1163 takami , m. , karr , j.l . , nisini , b. , et al . 2011 , apj , 743 , 193 takami , m. , karr , j. l. , koh , h. , et al . 2010 , apj , 720 , 155 teixeira , p.s . , mccoey , c. , fich , m. , & lada , c.j . 2008 , mnras , 384 , 71 velusamy , t. , langer , w.d . , & marsh , k.a . 2007 , apj , 668 , l159 werner , m.w . , roellig , t.l . , low , f.j . , et al . 2004 , apjs , 154 , 1 whitney , b. a. , wood , k. , bjorkman , j. e. , & cohen , m. 2003a , apj , 598 , 1079 whitney , b. a. , wood , k. , bjorkman , j. e. , & wolff , m. j. 2003b , apj , 591,1049 wouterloot , j.g.a . , & brand , j. 1999 , a&as , 140 , 177 wright , e. l. , eisenhardt , p. r. m. , mainzer , a. k. , et al . 2010 , aj , 140 , 1868 ybarra , j.e . , & lada , e.a . 2009 , apj , 695 , l120 ( yl09 ) ybarra , j. e. , lada , e. a. , balog , z. , et al . 2010 , apj , 714 , 469 below , we present a short description of both the identified jets and their candidate exciting sources . the numbering is that given in tables [ tab : tab1 ] and [ tab : tab3 ] . references to papers where possible associations are described ( e.g. sub - mm cores , blast sources , near - ir imaging ) are given only for _ jet 1 _ and no longer repeated . * _ jet 1 _ - extended bipolar jet composed of eight relatively bright knots . its morphology is reminiscent of a _ blue- and red - shifted jet components correspond to the nw ( knots 1 - 2 - 3 - 4 ) and se ( knots 5 - 6 - 7 - 8 ) lobe , respectively . it is located in a region dominated by the strong diffuse emission from a compact hii region ( 263.619 - 0.533 - caswell & haynes 1987 ) associated with an ir cluster ( massi et al . 2003 ) named irs16 following the nomenclature by liseau et al . the potential driving source emerges only in the irac bands up to 24@xmath1 m showing a sed with a classi spectral index . this source is also associated with a blast ( olmi et al . 2009 ) and a 1.2 mm dust core ( massi et al . 2007 ) . it is not aligned with the sequence of identified knots , but only with few of them of the nw lobe ( @xmath76 1,2,3,4 ) . therefore , we can not confirm this source as the jet es . see figure[jet_1:fig ] . * _ jet 2 _ - one of the nicest examples of very compact ( 0.13 pc ) bipolar jet with well defined blue- ( ne ) and red - shifted ( sw ) lobes bracketing a very young exciting source . this latter is point - like and detected only in the mips 24,70 @xmath1 m bands without any irac counterpart . given its very red [ 24]-[70 ] color of 7.4 , the system appears a suitable target to investigate the very early stages of the mass ejection ( and accretion ) process . an observational benefit is offered by its peculiar location rather isolated ( see figure[all : fig ] ) and unperturbed by excessive turbulence occurring in proximity of hii regions or ir clusters . remarkably , the source is associated with a blast source and with a compact dust core unresolved at 1.2 mm . see figure[jet_2:fig ] . * _ jet 3 _ - the morphology is very similar to that described for jet 2 : both show surprisingly the same pa orthogonal to the galactic magnetic field . it is composed of only two knots ( the blue - shifted one at ne and red - shifted one at sw ) aligned in a very compact ( 0.08 pc ) and collimated shape . both knots are symmetrical with respect to the position of the driving source which , again , is detected only in the mips bands with a [ 24]-[70 ] color of 6.0 . also this jet lies in a quiet portion of vmr - d . see figure[jet_3:fig ] . * _ jet 4 _ - it is one of the two longest ( 0.68 pc ) jets identified in the present survey , composed of the brightest detected knots . it preserves a high degree of collimation along all its length and shows bow - shock morphologies in both lobes oriented in opposite directions . blue- and red - shifted lobes ( northward and southward , composed by knots 1,2,3,4 and 5,6 respectively ) are emanated from a well identified ir cluster ( irs20 ) , where the source @xmath7698 ( massi et al . 1999 ) seems associated with a bipolar ir nebula oriented along the same jet axis . this source also coincides with a blast source and a resolved 1.2 mm dust core , with a spectral index typical of a classi source . the present observations allow us to reconsider our early suggestion coming from a lower sensitivity survey ( lorenzetti et al . 2002b ) , according to which the northern bow shock was associated with a faint ( unrelated ) star . see figure[jet_4:fig ] . * _ jet 5 _ - jet composed of well aligned knots . the membership of knot 1 to this jet is however doubtful , because this knot is around 170arcsec ( i.e. 0.57 pc ) distant from the others . two objects appear as possible candidate es : one ( 5a , to which we give higher priority ) located between knots 2 - 6 and 7 , in which case knot1 should be unrelated to jet5 , and one ( 5b ) between knot1 and all the others , in which case the jet should be parsec - scale . source 5a is barely detected with irac , while it is bright at 24@xmath1 m . source 5b is detected at all the wavelengths and it is associated both with a blast source and with an unresolved dust core . it shows a sed with a flat spectral index , which possibly signals it is not a very young object . see figure[jet_5:fig ] . * _ jet 6 _ - jet composed of five knots perfectly aligned with a candidate driving source north of them ; this latter is classified as class i object with a bright irac / mips counterpart coinciding with a 1.2 mm peak of dust emission ; it is also part of the young ir cluster irs19 ( liseau et al . according to this scenario the jet morphology results to be one - sided with its counter - jet possibly propagating northward deep inside the cloud . see figure[jet_6:fig ] . * _ jet 7 _ - jet discovered in this survey . it is composed of two irac knots well aligned with a candidate exciting source with a sed with flat spectral index and having valid fluxes up to 70@xmath1 m . the jet is apparently one - sided , displaying just a se lobe . see figure[jet_7:fig ] . * _ jet 8 _ - two knots aligned in positions symmetrically opposed with respect to the candidate es located at the system center . according to our analysis the red - shifted ( blue- ) lobe extends northward ( southward ) . the source displays a classi sed with valid detected counterparts from 1.6@xmath1 m to 70@xmath1 m ; it is also associated with both a blast source and a resolved 1.2 mm dust peak . see figure[jet_8:fig ] . * _ jet 9 _ - compact jet composed of five knots not perfectly aligned with the candidate es . this is classified as a class i source coinciding with a 1.2 mm dust core . see figure[jet_9:fig ] . * _ jet 10 _ - tentatively considered as a bipolar jet emanated by a young object ( whose sed is classified as classi ) . it is associated with both a blast source and an unresolved dust peak . such a candidate es is located at the jet center , but slightly displaced with respect to the knot axis . for this reason , a different origin of the observed knots can not be ruled out . see figure[jet_10:fig ] . * _ jet 11 _ - it appears as the most compact jet of the present survey ( 0.07 pc ) , composed of two knots , being the southeastern the blue- and the northwestern the red - shifted one . they seem symmetrically positioned at both sides of a faint candidate driving source detected only at 24@xmath1 m and without any association with cores at longer wavelengths . further observations are needed to better understand the true nature of this source . see figure[jet_11:fig ] . * _ jet 12 _ - three knots very close to those of _ jet 11 _ , but presenting a different alignment . blue- and red - shifted lobes are not straightforwardly recognizable , with some indication in favor of the blue ( red)-shifted lobe displaced ne ( sw ) . their position along a well defined axis is compatible with a candidate es located roughly in the central position . this source is associated with both a blast and 1.2 mm core . see figure[jet_12:fig ] . * _ jet 13 _ - together with _ jet 4 _ it appears as the longest jet of our survey . it is composed of six detected knots , of which knot 3 is practically coincident with the candidate es while knot 4 extends slightly displaced toward sw . while knots 3 - 4 represent the most compact part of the jet , pairs of knots 1 - 2 and 5 - 6 are located far away from the exciting sources , but still preserving the same good alignment . in particular , both pairs display a shape well associated to opposite bows : the overall jet morphology suggests the occurrence of two ( at least ) episodes of matter ejection and that the time elapsed between them has been rather long . at a good level of confidence , we consider the southern ( northern ) lobe as the blue-(red-)shifted one . the recognized exciting source is bright at 24@xmath1 m , has an extremely red mips color ( [ 24]-[70 ] = 7.9 ) , and is associated with a blast and 1.2 mm core . although it is a suitable candidate of a genuine protostellar object in a very early evolutionary stage , some doubts exist about its recognition as the jet es ( at least as far as the two extreme groups of knots is concerned ) . indeed , while the source appears as one of the coolest and youngest of our survey , the jet ( tentatively associated with it ) is definitely the longest : such an occurrence appears rather contradictory if we assume the [ 24]-[70 ] color as an age indicator . indeed , if the jet were composed only of knots 3 and 4 , its length should be only 0.11 pc . therefore , we prefer to not consider this jet in the analysis of figure[fig : jetcolor ] , which is based on the estimate of the projected jet length . see figure[jet_13:fig ] . * _ jet 14 _ - jet composed of six knots roughly aligned in e - w direction . one of them ( indicated as 14 - 1 ) could be alternatively attributed to the _ jet 15 _ ( knot 15 - 5 ) . two different candidate es have been found ( labeled as 14a and 14b ) : the former ( classified as a classi , to which we give higher priority ) belongs to an ir cluster and has a 2mass and a mips counterpart , while the latter is a very faint irac source ( without any wise confirmation ) with a flat spectral index . see figure[jet_14:fig ] . * _ jet 15 _ - we have presented this jet along with the candidate es in previous papers ( lorenzetti et . al 2002 ; giannini et al . 2005 ) to which the reader is referred to have a more complete view of that region . the correspondence of the knots identification given in those papers with that given here , is the following : the complex of knots previously identified as a1-a2-a3 , corresponds here to 15 - 1 , b1-b2 to 15 - 2 , g1 to g5 to 15 - 3 , i to 15 - 4 . knot 15 - 5 is identified here for the first time , but its connection with jet 15 or jet 14 is doubtful ( see also _ jet 14 _ ) . in the present paper other sparse knots ( k1 to k4 ) have been discovered in this region , but any association with a plausible es can not be provided . see figure[jet_15:fig ] . ccccc 1 & 8 & 89/0.30 & 163 & 1,2 + 2 & 4 & 34/0.13 & 57 & 1,2 + 3 & 2 & 24/0.08 & 57 & 1 + 4@xmath77 & 6 & 200/0.68 & 174 & 1,2,3 + 5@xmath78 & 6(7 ) & 106/0.36 ( 321/1.09 ) & 48 ( 54 ) & 1,2 + 6@xmath79 & 5 & 74@xmath80/0.25@xmath80 & 178 & 1,2,3 + 7 & 2 & 38/0.14 & 121 & - + 8 & 2 & 112/0.37 & 5 & 2 + 9 & 5 & 56/0.19 & 5 & 2 + 10 & 2 & 106/0.36 & 43 & 2 + 11 & 2 & 20/0.07 & 146 & 2 + 12 & 3 & 86/0.29 & 112 & 2 + 13@xmath78 & 6(2 ) & 206/0.70 ( 30/0.11 ) & 173 & 2 + 14@xmath78 & 6(5 ) & 155/0.53 ( 76/0.26 ) & 94 ( 96 ) & - + 15@xmath81 & 4(5 ) & 128/0.44 ( 139/0.48 ) & 155 ( 156 ) & 3,4 + ccccccccccc 1 - 1 & 08:45:33.8 & -43:51:24 & 15 & 0.171 @xmath82 0.001 & @xmath440.3 & 2.73 @xmath82 0.09 & @xmath443 & @xmath443 & 8.217 & 0.052 + 1 - 2 & 08:45:34.1 & -43:51:28 & 32 & 0.299 @xmath82 0.002 & @xmath440.9 & 3.10 @xmath82 0.09 & @xmath441 & @xmath441 & 9.330 & 0.091 + 1 - 3 & 08:45:34.2 & -43:51:41 & 9 & 0.130 @xmath82 0.002 & 0.7 @xmath82 0.2 & 0.84 @xmath82 0.04 & @xmath443 & @xmath443 & 2.740 & 0.040 + 1 - 4 & 08:45:34.4 & -43:51:47 & 11 & 0.105 @xmath82 0.002 & @xmath440.2 & 0.19 @xmath82 0.06 & @xmath442 & @xmath443 & 0.557 & 0.032 + 1 - 5 & 08:45:35.2 & -43:52:00 & 24 & 0.074 @xmath82 0.002 & 0.52 @xmath82 0.07 & 0.5 @xmath82 0.1 & @xmath443 & @xmath444 & 1.739 & 0.023 + 1 - 6 & 08:45:35.4 & -43:52:12 & 28 & 0.185 @xmath82 0.003 & 0.98 @xmath82 0.05 & 0.93 @xmath82 0.03 & @xmath442 & @xmath446 & 3.076 & 0.056 + 1 - 7 & 08:45:35.7 & -43:52:22 & 14 & 0.090 @xmath82 0.002 & 0.48 @xmath82 0.03 & 1.23 @xmath82 0.04 & 3.8 @xmath82 0.2 & @xmath442 & 13.127 & 0.028 + 1 - 8 & 08:45:35.6 & -43:52:36 & 20 & 0.161 @xmath82 0.002 & 0.74 @xmath82 0.03 & 1.17 @xmath82 0.03 & 3.7 @xmath82 0.3 & @xmath443 & 12.797 & 0.049 + 2 - 1 & 08:48:16.8 & -43:47:08 & 46 & 0.129 @xmath82 0.002 & 0.28 @xmath82 0.01 & 0.66 @xmath82 0.01 & 0.21 @xmath82 0.03 & @xmath441 & 2.583 & 0.040 + 2 - 2 & 08:48:16.0 & -43:47:14 & 15 & 0.025 @xmath82 0.002 & @xmath440.03 & 0.31 @xmath82 0.01 & @xmath440.2 & @xmath443 & 0.001 & 0.008 + 2 - 3 & 08:48:15.4 & -43:47:17 & 12 & 0.034 @xmath82 0.002 & 0.12 @xmath82 0.01 & 0.14 @xmath82 0.01 & 0.05 @xmath82 0.01 & @xmath442 & 0.561 & 0.010 + 2 - 4 & 08:48:14.2 & -43:47:26 & 20 & 0.066 @xmath82 0.002 & 0.14 @xmath82 0.01 & 0.34 @xmath82 0.01 & 0.93 @xmath82 0.02 & @xmath441 & 3.332 & 0.020 + 3 - 1 & 08:46:34.9 & -43:21:07 & 18 & 0.069 @xmath82 0.002 & 0.19 @xmath82 0.01 & 0.76 @xmath82 0.01 & 0.54 @xmath82 0.02 & 0.7 @xmath82 0.02 & 5.482 & 0.021 + 3 - 2 & 08:46:34.0 & -43:21:16 & 21 & 0.055 @xmath82 0.003 & 0.22 @xmath82 0.01 & 0.67 @xmath82 0.04 & 1.20 @xmath82 0.03 & 0.7 @xmath82 0.04 & 6.953 & 0.017 + 4 - 1 & 08:49:25.8 & -43:15:41 & 15 & 0.076 @xmath82 0.001 & 0.19 @xmath82 0.01 & 0.40 @xmath82 0.01 & 0.4 @xmath82 0.1 & @xmath440.6 & 2.239 & 0.023 + 4 - 2 & 08:49:24.8 & -43:15:50 & 120 & 0.839 @xmath82 0.004 & 2.95 @xmath82 0.07 & 4.98 @xmath82 0.03 & 8.2 @xmath82 0.2 & @xmath442 & 35.798 & 0.256 + 4 - 3 & 08:49:25.6 & -43:15:54 & 60 & 0.894 @xmath82 0.004 & 2.65 @xmath82 0.03 & 4.09 @xmath82 0.04 & 5.7 @xmath82 0.1 & @xmath441 & 26.866 & 0.273 + 4 - 4 & 08:49:25.4 & -43:16:17 & 85 & 0.984 @xmath82 0.004 & 2.39 @xmath82 0.06 & 5.33 @xmath82 0.07 & 15.7 @xmath82 0.4 & @xmath443 & 55.056 & 0.301 + 4 - 5 & 08:49:26.6 & -43:18:02 & 370 & 1.044 @xmath82 0.008 & 1.3 @xmath82 0.1 & 6.1 @xmath82 0.1 & 22.0 @xmath82 0.7 & @xmath446 & 72.709 & 0.319 + 4 - 6 & 08:49:27.2 & -43:18:58 & 70 & 0.165 @xmath82 0.004 & 0.35 @xmath82 0.01 & 1.18 @xmath82 0.01 & @xmath440.3 & @xmath440.6 & 3.649 & 0.050 + 5 - 1 & 08.48.42.4 & -43.29.43 & 12 & 0.040 @xmath82 0.002 & 0.12 @xmath82 0.01 & 0.18 @xmath82 0.01 & 0.26 @xmath82 0.03 & 0.43 @xmath82 0.03 & 2.369 & 0.012 + 5 - 2 & 08:48:26.2 & -43:31:39 & 12 & 0.231 @xmath82 0.003 & 0.21 @xmath82 0.01 & 0.95 @xmath82 0.01 & 0.93 @xmath82 0.08 & @xmath440.2 & 5.182 & 0.071 + 5 - 3 & 08:48:25.4 & -43:31:44 & 44 & 0.085 @xmath82 0.004 & 0.08 @xmath82 0.01 & 1.24 @xmath82 0.02 & 0.13 @xmath82 0.03 & @xmath440.3 & 4.151 & 0.026 + 5 - 4 & 08:48:23.6 & -43:31:55 & 15 & 0.137 @xmath82 0.003 & 0.37 @xmath82 0.01 & 0.70 @xmath82 0.02 & 0.62 @xmath82 0.06 & @xmath440.1 & 3.736 & 0.042 + 5 - 5 & 08:48:22.7 & -43:32:04 & 24 & 0.214 @xmath82 0.003 & 0.63 @xmath82 0.01 & 1.45 @xmath82 0.01 & 2.41 @xmath82 0.08 & 0.9 @xmath82 0.2 & 12.710 & 0.065 + 5 - 6 & 08:48:22.4 & -43:32:00 & 9 & 0.002 @xmath82 0.002 & 0.09 @xmath82 0.01 & 0.13 @xmath82 0.01 & @xmath440.6 & @xmath440.1 & 0.405 & 0.001 + 5 - 7 & 08:48:18.9 & -43:32:49 & 40 & 0.011 @xmath82 0.003 & 0.04 @xmath82 0.01 & 0.15 @xmath82 0.01 & @xmath440.6 & @xmath440.3 & 0.462 & 0.004 + 6 - 1 & 08:48:53.2 & -43:31:10 & 9 & 0.027 @xmath82 0.003 & 0.04 @xmath82 0.01 & 0.25 @xmath82 0.01 & 0.37 @xmath82 0.03 & @xmath442 & 1.661 & 0.008 + 6 - 2 & 08:48:53.3 & -43:31:24 & 12 & 0.034 @xmath82 0.002 & 0.16 @xmath82 0.01 & 0.44 @xmath82 0.09 & 0.37 @xmath82 0.03 & @xmath442 & 2.276 & 0.010 + 6 - 3 & 08:48:53.3 & -43:31:34 & 16 & 0.035 @xmath82 0.002 & 0.03 @xmath82 0.01 & 0.14 @xmath82 0.01 & @xmath440.2 & @xmath442 & 0.440 & 0.011 + 6 - 4 & 08:48:53.3 & -43:31:49 & 22 & 0.104 @xmath82 0.003 & 0.52 @xmath82 0.01 & 0.98 @xmath82 0.02 & 0.46 @xmath82 0.04 & @xmath443 & 4.216 & 0.032 + 6 - 5 & 08.48.53.3 & -43:32:12 & 55 & 0.374 @xmath82 0.003 & 0.95 @xmath82 0.01 & 2.90 @xmath82 0.03 & 4.2 @xmath82 0.1 & 3.3 @xmath82 0.1 & 27.648 & 0.114 + 7 - 1 & 08:48:45.9 & -43:16:14 & 24 & - & 0.09 @xmath82 0.01 & 0.14 @xmath82 0.01 & @xmath440.3 & @xmath440.3 & 0.434 & - + 7 - 2 & 08:48:47.4 & -43:16:27 & 32 & - & 0.28 @xmath82 0.01 & 0.65 @xmath82 0.01 & 0.77 @xmath82 0.03 & 0.6 @xmath82 0.1 & 5.459 & - + 8 - 1 & 08:49:13.8 & -43:35:42 & 35 & 0.237 @xmath82 0.003 & 0.45 @xmath82 0.01 & 1.31 @xmath82 0.03 & 0.9 @xmath82 0.2 & 0.6 @xmath82 0.2 & 7.701 & 0.072 + 8 - 2 & 08:49:13.0 & -43:37:22 & 51 & 0.030 @xmath82 0.004 & 0.41 @xmath82 0.01 & 0.32 @xmath82 0.01 & @xmath440.2 & @xmath440.3 & 1.092 & 0.023 + 9 - 1 & 08:48:59.4 & -43:37:46 & 44 & 0.124 @xmath82 0.004 & 0.42 @xmath82 0.02 & 0.53 @xmath82 0.01 & @xmath440.3 & @xmath440.2 & 1.722 & 0.038 + 9 - 2 & 08:48:59.2 & -43:38:01 & 36 & 0.027 @xmath82 0.004 & 0.14 @xmath82 0.01 & 0.27 @xmath82 0.01 & @xmath440.1 & @xmath440.2 & 0.837 & 0.008 + 9 - 3 & 08:48:58.8 & -43:38:12 & 18 & 0.031 @xmath82 0.003 & 0.02 @xmath82 0.01 & 0.37 @xmath82 0.01 & @xmath440.1 & @xmath440.2 & 1.127 & 0.010 + 9 - 4 & 08:48:58.5 & -43:38:24 & 38 & 0.194 @xmath82 0.003 & 0.50 @xmath82 0.02 & 2.10 @xmath82 0.01 & 2.98 @xmath82 0.08 & @xmath440.3 & 13.759 & 0.003 + 9 - 5 & 08:48:58.6 & -43:38:33 & 22 & 0.095 @xmath82 0.002 & 0.45 @xmath82 0.01 & 0.54 @xmath82 0.01 & 0.41 @xmath82 0.05 & @xmath440.3 & 2.788 & 0.029 + 10 - 1 & 08:47:45.5 & -43:42:57 & 20 & 0.089 @xmath82 0.001 & 0.23 @xmath82 0.01 & 0.69 @xmath82 0.01 & 0.74 @xmath82 0.03 & 0.6 @xmath82 0.1 & 5.563 & 0.027 + 10 - 2 & 08:47:38.7 & -43:44:13 & 30 & 0.079 @xmath82 0.002 & 0.41 @xmath82 0.01 & 0.38 @xmath82 0.01 & 1.35 @xmath82 0.03 & 0.3 @xmath82 0.1 & 5.468 & 0.024 + 11 - 1 & 08:46:56.4 & -43:52:42 & 13 & 0.021 @xmath82 0.001 & 0.18 @xmath82 0.01 & 0.51 @xmath82 0.01 & 0.42 @xmath82 0.02 & @xmath440.1 & 2.617 & 0.006 + 11 - 2 & 08:46:57.3 & -43:52:58 & 22 & 0.018 @xmath82 0.003 & 0.04 @xmath82 0.01 & 0.14 @xmath82 0.01 & @xmath440.06 & @xmath440.9 & 0.428 & 0.005 + 12 - 1 & 08:46:47.3 & -43:52:44 & 35 & 0.137 @xmath82 0.003 & 0.26 @xmath82 0.01 & 0.74 @xmath82 0.01 & 0.69 @xmath82 0.02 & 0.37 @xmath82 0.04 & 4.918 & 0.042 + 12 - 2 & 08:46:51.1 & -43:53:08 & 30 & 0.044 @xmath82 0.003 & 0.14 @xmath82 0.01 & 0.20 @xmath82 0.01 & @xmath440.6 & 0.32 @xmath82 0.04 & 1.469 & 0.013 + 12 - 3 & 08:46:53.8 & -43:53:15 & 36 & 0.097 @xmath82 0.003 & 0.34 @xmath82 0.01 & 0.58 @xmath82 0.01 & 0.62 @xmath82 0.01 & 0.80 @xmath82 0.04 & 5.440 & 0.030 + 13 - 1@xmath79 & 08:49:28.0 & -44:03:13 & 56 & 0.122 @xmath82 0.004 & 0.45 @xmath82 0.01 & 1.04 @xmath82 0.01 & 0.87 @xmath82 0.03 & @xmath440.1 & 5.377 & 0.037 + 13 - 2@xmath79 & 08:49:27.3 & -44:03:20 & 90 & 0.343 @xmath82 0.004 & 0.61 @xmath82 0.01 & 1.36 @xmath82 0.01 & 1.13 @xmath82 0.04 & 0.50 @xmath82 0.05 & 8.321 & 0.105 + 13 - 3@xmath80 & 08:49:28.7 & -44:04:29 & 50 & 0.046 @xmath82 0.004 & 0.30 @xmath82 0.01 & 1.19 @xmath82 0.01 & 0.48 @xmath82 0.03 & 1.09 @xmath82 0.01 & 7.679 & 0.014 + 13 - 4 & 08:49:29.4 & -44:04:53 & 45 & 0.097 @xmath82 0.004 & 0.34 @xmath82 0.01 & 0.56 @xmath82 0.08 & @xmath440.1 & @xmath440.2 & 1.784 & 0.030 + 13 - 5@xmath79 & 08:49:28.2 & -44:06:28 & 15 & 0.027 @xmath82 0.004 & 0.11 @xmath82 0.01 & 0.18 @xmath82 0.01 & @xmath440.1 & @xmath440.1 & 0.572 & 0.008 + 13 - 6@xmath79 & 08:49:29.5 & -44:06:34 & 20 & 0.018 @xmath82 0.001 & 0.04 @xmath82 0.01 & 0.14 @xmath82 0.01 & @xmath440.1 & @xmath440.1 & 0.424 & 0.006 + 14 - 1@xmath83 & 08:46:36.0 & -43:56:02 & 36 & - & 0.13 @xmath82 0.01 & 0.24 @xmath82 0.01 & @xmath440.2 & 0.7 @xmath82 0.3 & 2.672 & - + 14 - 2 & 08:46:28.7 & -43:55:57 & 24 & - & 0.10 @xmath82 0.02 & 0.29 @xmath82 0.01 & 0.63 @xmath82 0.04 & @xmath440.2 & 2.429 & - + 14 - 3 & 08:46:27.0 & -43:55:52 & 30 & - & 0.27 @xmath82 0.01 & 1.04 @xmath82 0.01 & 0.73 @xmath82 0.03 & 0.60 @xmath82 0.06 & 6.578 & - + 14 - 4 & 08:46:24.3 & -43:55:45 & 15 & - & 0.05 @xmath82 0.01 & 0.10 @xmath82 0.01 & 0.14 @xmath82 0.02 & @xmath440.1 & 0.651 & - + 14 - 5 & 08:46:21.8 & -43:55:46 & 45 & - & 0.48 @xmath82 0.01 & 1.10 @xmath82 0.08 & 0.75 @xmath82 0.04 & @xmath440.3 & 5.285 & - + 14 - 6 & 08:46:21.7 & -43:55:56 & 10 & - & 0.07 @xmath82 0.01 & 0.26 @xmath82 0.01 & 0.11 @xmath82 0.01 & 0.16 @xmath82 0.02 & 1.512 & - + 15 - 1 & 08:46:30.9 & -43:53:18 & 60 & - & 0.81 @xmath82 0.02 & 1.95 @xmath82 0.01 & 2.6 @xmath82 0.2 & 6.8 @xmath82 0.4 & 30.121 & 0.101 + 15 - 2 & 08:46:31.8 & -43:53:51 & 65 & - & 0.42 @xmath82 0.01 & 1.06 @xmath82 0.01 & 1.5 @xmath82 0.1 & 0.5 @xmath82 0.3 & 8.230 & 0.039 + 15 - 3 & 08:46:35.4 & -43:55:08 & 110 & - & 1.82 @xmath82 0.02 & 5.96 @xmath82 0.03 & 2.5 @xmath82 0.5 & 9.9 @xmath82 0.1 & 50.312 & 0.186 + 15 - 4 & 08:46:35.7 & -43:55:22 & 25 & - & 0.10 @xmath82 0.01 & 0.13 @xmath82 0.02 & 0.7 @xmath82 0.1 & 1.1 @xmath82 0.1 & 4.902 & 0.014 + 15 - 5@xmath83 & 08:46:36.0 & -43:56:02 & 36 & - & 0.13 @xmath82 0.01 & 0.24 @xmath82 0.01 & @xmath440.2 & 0.7 @xmath82 0.3 & 2.672 & - + k-1 & 08:46:30.1 & -43:54:11 & 18 & - & 0.22 @xmath82 0.01 & 0.46 @xmath82 0.04 & 0.34 @xmath82 0.05 & 1.07 @xmath82 0.06 & 5.060 & - + k-2 & 08:46:27.5 & -43:54:15 & 28 & - & 0.05 @xmath82 0.01 & 0.14 @xmath82 0.01 & @xmath440.06 & 0.18 @xmath82 0.04 & 0.886 & - + k-3 & 08:46:29.1 & -43:54:40 & 20 & - & 0.07 @xmath82 0.01 & 0.26 @xmath82 0.01 & 0.20 @xmath82 0.04 & @xmath440.1 & 1.291 & - + k-4 & 08:46:29.7 & -43:54:49 & 19 & - & @xmath440.02 & 0.10 @xmath82 0.01 & 0.07 @xmath82 0.02 & 0.35 @xmath82 0.06 & 1.380 & - + cccccccccccccc 1 & 08 45 34.5 & -43 51 57.3 & 18064 & - & j084534.45 - 435157.4 & 141 & j084535 - 435156 & mms2 & - & 0.5 & 1.0 & 7.0 & 7.5 + 2 & 08 48 15.9 & -43 47 15.6 & - & - & - & 657 & j084815 - 434714 & umms16 & - & - & - & 4.2 & 8.1 + 3 & 08 46 34.3 & -43 21 11.2 & - & - & - & 298 & - & - & - & - & - & - & - + 4 & 08 49 26.3 & -43 17 10.8 & 126339 & 08492636 - 4317125 & j084926.28 - 431711.1 & 991 & j084925 - 431710 & mms22 & 1.9 & 0.4 & 1.6 & 5.0 & 12.3 + 5a & 08 48 20.5 & -43 32 29.1 & 95972 & - & - & 678 & - & - & - & - & 1.1 & - & - + 5b & 08 48 33.9 & -43 30 47.3 & 103109 & 08483394 - 433047 & j084833.95 - 433047.3 & 729 & j084833 - 433056 & umms19 & 0.5 & 0.6 & 0.6 & 9.2 & 9.3 + 6 & 08 48 53.2 & -43 30 55.2 & 112469 & - & j084853.17 - 433055.0 & 838 & - & mms16 & - & 1.9 & 2.1 & - & 5.9 + 7 & 08 48 44.4 & -43 16 8.06 & 108306 & - & j084844.38 - 431607.5 & 787 & - & - & - & 0.5 & 1.3 & - & - + 8 & 08 49 13.0 & -43 36 24.5 & 120963 & 08491298 - 4336257 & j084913.19 - 433623.3 & 928 & j084912 - 433618 & mms21 & 1.2 & 2.4 & 1.8 & 9.2 & 4.2 + 9 & 08 48 58.7 & -43 38 21.1 & 114979 & - & - & 858 & - & mms17 & - & - & 0.6 & - & 12.0 + 10 & 08 47 42.8 & -43 43 48.4 & 74548 & 08474283 - 4343482 & j084742.80 - 434348.3 & 546 & j084742 - 434347 & umms11 & 0.4 & 0.1 & 0.4 & 5.6 & 9.8 + 11 & 08 46 56.8 & -43 52 52.2 & - & - & - & 392 & - & - & - & - & 1.5 & - & - + 12 & 08 46 49.1 & -43 52 54.3 & 48370 & - & - & 355 & j084648 - 435257 & mms5 & - & - & 0.9 & 6.2 & 14.3 + 13 & 08 49 28.7 & -44 04 29.2 & - & - & j084928.69 - 440429.2 & 998 & j084928 - 440426 & mms23 & - & 1.0 & 1.2 & 7.1 & 10.5 + 14a & 08 46 31.4 & -43 55 56.9 & 41256 & 08463136 - 4355569 & j084631.35 - 435557.1 & 285 & - & - & 0.0 & 0.5 & 9.5 & - & - + 14b & 08 46 25.3 & -43 55 44.8 & 38798 & - & - & - & - & - & - & - & - & - & - + ccccccccccccc 1 & - & - & - & [email protected] & [email protected] & 40@xmath821 & 67@xmath822 & 128@xmath827 & 905@xmath8287 & @xmath448886 & 3700 & 1.7 + 2 & - & - & - & - & - & - & - & - & [email protected] & 2469@xmath8236 & 60 & - + 3 & - & - & - & - & - & - & - & - & [email protected] & 343@xmath8217 & - & - + 4 & @xmath441.0 & @xmath448.5 & [email protected] & 152@xmath824 & 552@xmath8215 & 1496@xmath8230 & 2900@xmath8295 & 6221@xmath8280 & @xmath434000 ( 34884@xmath82257 ) & 48580@xmath821566 & 1320 & 3.2 + 5a & - & - & - & [email protected] & [email protected] & @xmath440.3 & @xmath440.4 & - & [email protected] & @xmath44180 & - & - + 5b & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 29@xmath822 & 198@xmath825 & 1014@xmath8221 & 110 & 0.3 + 6 & - & - & - & [email protected] & [email protected] & [email protected] & 70@xmath821 & 86@xmath822 & 1090@xmath8245 & 4867@xmath8254 & 230 & 1.8 + 7 & - & - & - & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 90@xmath823 & 968@xmath8213 & - & 0.3 + 8 & @xmath441.8 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 53@xmath821 & 583@xmath8265 & 3648@xmath82232 & 1760 & 0.8 + 9 & - & - & - & [email protected] & [email protected] & [email protected] & [email protected] & - & 113@xmath828 & 1175@xmath8256 & 760 & 1.3 + 10 & @xmath441.0 & @xmath4410.3 & [email protected] & 42@xmath822 & 79@xmath822 & 124@xmath823 & 180@xmath824 & 332@xmath823 & 1163@xmath8280 & 3519@xmath8239 & 100 & 1.1 + 11 & - & - & - & - & - & - & - & - & [email protected] & @xmath44144 & - & - + 12 & - & - & - & @xmath44 0.2 & [email protected] & [email protected] & [email protected] & - & 32.9@xmath821 & @xmath44390 & 280 & - + 13 & - & - & - & - & - & - & - & - & [email protected] & 811@xmath8219 & 140 & - + 14a & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 30@xmath822 & 271@xmath824 & @xmath434000 ( 5294@xmath82102 ) & 11870@xmath82645 & - & 0.7 + 14b & - & - & - & [email protected] & [email protected] & [email protected] & [email protected] & - & @xmath84 & @xmath85 & - & 0.2 + cccc 1 & 1 - 2 - 3 - 4 & 5 - 6 - 7 - 8 & 6 + 2 & 1 & 4 & 11 + 3 & 1 & 2 & 5 + 4 & 1 - 2 - 3 - 4 & 5 & 6 + 5a@xmath77 & 7 & 2 - 3 - 4 - 5 - 6 & 2 + 5b@xmath77 & 2 - 3 - 4 - 5 - 6 - 7 & 1 & 2 + 8 & 2 & 1 & 8 + 9 & 5 & 1 - 2 - 3 & 2 + 10 & 2 & 1 & 0 + 11 & 2 & 1 & 6 + 12 & 1 & 2 & 4 + 13a@xmath78 & 5 - 6 & 1 - 2 & 4 + 13a@xmath78 & 4 & 3 & 10 + 15 & 1 - 2 & 3 - 4 & 2 + cccccccccc 1 & * 1000 * ( 758 ) & * 20 * ( 40 ) & * 17 * ( 21 ) & * 738 * ( 1050 ) & * 3923 * ( 3865 ) & * 0.87 * ( 0.74 ) & * 15.8 * ( 13.2 ) & * 43 * ( 47 ) & * 19 * ( 13.7 ) + 2@xmath77 & 500 - 1000 & - & - & - & 51 & - & - & - & 0.3 - 1.3 + 3@xmath77 & 500 - 1000 & - & - & - & 47 & - & - & - & 0.2 - 0.8 + 4 & * 692 * ( 660 ) & * 55 * ( 65 ) & * 18 * ( 24 ) & * 107 * ( 876 ) & * 4162 * ( 4162 ) & * 1.98 * ( 1.98 ) & * 2.6 * ( 2.6 ) & * 0.08 * ( 0.07 ) & * 216 * ( 216 ) + 5b & * 759 * ( 832 ) & * 20 * ( 20 ) & * 11 * ( 7 ) & * 23 * ( 24 ) & * 2887 * ( 3003 ) & * 0.15 * ( 0.18 ) & * 1.3 * ( 3.0 ) & * 0.05 * ( 0.14 ) & * 1.1 * ( 1.6 ) + 6 & * 692 * ( 912 ) & * 50 * ( 40 ) & * 61 * ( 71 ) & * 37 * ( 28 ) & * 3535 * ( 3160 ) & * 0.40 * ( 0.23 ) & * 6.9 * ( 2.2 ) & * 0.16 * ( 0.03 ) & * 16 * ( 30 ) + 7 & * 759 * ( 501 ) & * 30 * ( 50 ) & * 43 * ( 45 ) & * 26 * ( 25 ) & * 3039 * ( 3283 ) & * 0.20 * ( 0.27 ) & * 1.1@xmath78 & 0.05@xmath78 & 1.4 * ( 0.8 ) + 8 & * 759 * ( 692 ) & * 20 * ( 20 ) & * 24 * ( 21 ) & * 22 * ( 80 ) & * 4370 * ( 3949 ) & * 0.35 * ( 0.68 ) & * 7.0 * ( 9.1 ) & * 3.5 * ( 8.3 ) & * 4.7 * ( 5.6 ) + 9 & * 692 * ( 1000 ) & * 20 * ( 20 ) & * 57 * ( 67 ) & * 94 * ( 2 ) & * 3040 * ( 3620 ) & * 0.19 * ( 0.42 ) & * 3.3 * ( 3.8 ) & * 2.7 * ( 3.5 ) & * 1.3 * ( 2.8 ) + 10 & * 955 * ( 871 ) & * 20 * ( 40 ) & * 16 * ( 7 ) & * 8 * ( 29 ) & * 2970 * ( 3787 ) & * 0.18 * ( 0.65 ) & * 3.8 * ( 7.9 ) & * 0.2 * ( 0.3 ) & * 12 * ( 14 ) + 13@xmath77 & 500 - 1000 & - & - & - & 43 & - & - & - & 0.2 - 0.9 + 14a & * 955 * ( 661 ) & * 90 * ( 90 ) & * 10 * ( 5 ) & * @xmath43 1000 * ( @xmath43 1000 ) & * 4386 * ( 4503 ) & * 3.17(2.63 ) & 5.7@xmath78 & 0.18@xmath78 & 109 * ( 38 ) + co(3 - 2 ) map of jet # 3 superposed on the irac three - color image . contours are ( from right to left ) 1.2 , 1,8 , 2.4 , 3.0 and 3.6 k km s@xmath12 for the red - lobe and 0.6 , 1.2 , 1,8 , 2.4 , 3.0 and 3.6 k km s@xmath12 ( from left to right ) for the blue - lobe , respectively . the velocity is integrated between -3.6 km s@xmath12 and + 0.8 km s@xmath12 for the blue - wing , and between + 0.8 km s@xmath12 and + 3.4 km s@xmath12 for the right - wing . the positions of the exciting source and the h@xmath0 knots are evidenced as well ( ellipse and boxes , respectively).[fig : apex],width=453 ] ; in particular the value associated with the best - fit model is indicated with a dot . for source # 13 , the vertical line indicates @xmath38 in case the jet is composed alternatively of two or six knots ( see appendix[sec : appendix]).[fig : lumlum],width=453 ]	we present a survey of h@xmath0 jets from young protostars in the vela - d molecular cloud ( vmr - d ) , based on _ spitzer_-irac data between 3.6@xmath1 m and 8.0@xmath1 m . our search has led to the identification of 15 jets ( 2 new discoveries ) and about 70 well aligned knots within 1.2 deg@xmath2 . we compare the irac maps with observations of the h@xmath0 1 - 0 s(1 ) line at 2.12@xmath1 m , with a _ spitzer_-mips map at 24@xmath1 m and 70@xmath1 m , and with a map of the dust continuum emission at 1.2 mm . from such a comparison we find a tight association between molecular jets and dust peaks . + the jet candidate exciting sources have been searched for in the published catalog of the young stellar objects of vmr - d . in particular , we searched for all the sources of class ii or ( preferentially ) earlier which are located close to the jet center and aligned with it . furthermore , the association between jet and exciting source was validated by estimating the differential extinction between the jet opposite lobes . we are able to find a best - candidate exciting source in all but two jets , for which two alternative candidates are given . four exciting sources are not ( or very barely ) observed at wavelengths shorter than 24@xmath1 m , suggesting they are very young protostars . three of them are also associated with the most compact jets ( projected length @xmath3 0.1 pc ) . + the exciting source spectral energy distributions ( seds ) have been constructed and modeled by means of all the available photometric data between 1.2@xmath1 m and 1.2 mm . from seds fits we derive the main source parameters , which indicate that most of them are low - mass protostars . + a significant correlation is found between the projected jet length and the [ 24 ] - [ 70 ] color , which is consistent with an evolutionary scenario according to which shorter jets are associated with younger sources . a rough correlation is found between irac line cooling and exciting source bolometric luminosity , in agreement with the previous literature . the emerging trend suggests that mass loss and mass accretion are tightly related phenomena and that both decrease with time .
ru - min wang wish to thank alexander lenz for useful suggestions . the work is supported by national natural science foundation of china under contract nos.10305003 , 10475031 , 10440420018 and by moe of china under projects ncet-04 - 0744 , srfdp-20040511005 , cfkstip-704035 . 100 a. abulencia et al . ( cdf collaboration ) , hep - ex/0606027 . v. m. abazov et al . ( d collaboration ) , hep - ex/0603029 . j. charles et al . ( ckmfitter group ) , eur . j. * c41 * , 1(2005 ) , hep - ph/0406184 , updated results and plots available at : @xmath86 . m. bona et al . ( utfit collaboration ) , jhep * 0603 * , 080(2006 ) , hep - ph/0509219 . s. hashimoto , int . * a20 * , 5133(2005 ) . m. blanke , a. j. buras , d. guadagnoli and c. tarantino , hep - ph/0604057 ; p. ball and r. fleischer , hep - ph/0604249 ; z. ligeti , m. papucci and g. perez , hep - ph/0604112 ; y. grossman , y. nir and g. raz , hep - ph/0605028 ; m. bona et al . ( utfit collaboration ) , hep - ph/0605213 . m. ciuchini and l. silvestrini , hep - ph/0603114 ; j. foster , k. i. okumura and l. roszkowski , hep - ph/0604121 ; s. khalil , hep - ph/0605021 ; m. endo and s. mishima , hep - ph/0603251 ; s. baek , hep - ph/0605182 ; g. isidori and p. paradisi , phys b639 * , 499(2006 ) . k. cheung et al . , hep - ph/0604223 ; xiao - gang he and g. valencia , hep - ph/0605202 ; s. baek , jong hun jeon and c. s. kim , hep - ph/0607113 . b. dutta and y. mimura , hep - ph/0607147 ; j. k. parry , hep - ph/0606150 ; j. k. parry , hep - ph/0608192 . s. nandi and j. p. saha , hep - ph/0608341 ; j. p. saha and a. kundu , phys . rev . * d69 * , 016004(2004 ) ; a. kundu and j. p. saha , phys . rev . * d70 * , 096002(2004 ) ; g. bhattacharyya and a. raychaudhuri , phys . rev . * d57 * , 3837(1998 ) ; d. choudhury and probir roy , phys . lett . * b378 * , 153(1996 ) ; k. agashe and m. graesser , phys . rev . * d54 * , 4445(1996 ) . b. de carlos and p. white , phys . rev . * d55 * , 4222(1997 ) ; d. guetta , phys . rev . * d58 * , 116008(1998 ) . g. buchalla , a. j. buras and m. e. lauteubacher , rev . phys . * 68 * , 1125(1996 ) . m. beneke et al . , phys . * b459 * , 631(1999 ) ; a. lenz , hep - ph/9906317 . m. beneke , g. buchalla and i. dunietz , phys . rev . * d54 * , 4419(1996 ) . a. lenz , hep - ph/0412007 . s. weinberg , phys . * d26 * , 287(1982 ) . zhi - zhong xing , eur . j. * c4 * , 283(1998 ) . y. grossman , phys . b380 * , 99(1996 ) . i. i. bigi , v. a. khoze , n. g. uraltsev and a. i. sanda , in cp violation , edited by c. jarlskog ( world scientific , singapore , 1988 ) , p.175 ; j. l. hewett , t. takeuchi and s. thomas , slac - pub-7088 or cern - th/96 - 56 ; y. grossman , y. nir , and r. rattazzi , slac - pub-7379 or cern - th-96 - 368 ; m. gronau and d. london , phys . rev . * d55 * , 2845(1997 ) . w. m. yao et al . , journal of physics * g33 * , 1(2006 ) . a. j. buras , m. jamin and p. h. weisz , nucl . phys . * b347 * , 491(1990 ) ; j. urban et al . , nucl . phys . * b523 * , 40(1998 ) . d. be@xmath87irevi@xmath87 et al . , j. high energy physics * 0204 * , 025(2002 ) . r. barbier et al . , phys . rept . * 420 * , 1(2005 ) ; m. chemtob , prog . part . . phys . * 54 * , 71(2005 ) . m. frank and h. hamidian , j. phys . * g24 * , 2203(1998 ) . b. c. allanach , a. dedes and h. k. dreiner , phys . * d69 * , 115002(2004 ) . v. m. abazov et al . ( d collaboration ) , phys . lett . * 95 * , 171801(2005 ) ; d. acosta et al . ( cdf collaboration ) , phys . . lett . * 94 * , 101803(2005 ) ; r. barate et al . ( aleph collaboration ) , phys . b486 * , 286(2000 ) .	the recent measurements of the @xmath0 mass difference @xmath1 by the cdf and d collaborations are roughly consistent with the standard model predictions , therefore , these measurements will afford an opportunity to constrain new physics scenarios beyond the standard model . we consider the impact of the r - parity violating supersymmetry in the @xmath2 mixing , and use the latest experimental results of @xmath1 to constrain the size of the r - parity violating tree level couplings in the @xmath2 mixing . then , using the constrained r - parity violating parameter space from @xmath1 , we show the r - parity violating effects on the @xmath0 width difference @xmath3 . * pacs numbers : 12.60.jv , 11.30.er , 12.15.mm , 14.40.nd * recently cdf and d collaborations have measured the mass difference in the @xmath2 system @xcite with the results @xmath4 the measurement of cdf collaboration turned out to be surprisingly below the standard model ( sm ) predictions obtained from other constraints @xcite @xmath5 a consistent though slightly smaller value is found for the mass difference directly from its sm expression in later eq.([smdms ] ) @xmath6 with the input parameters collected in table i. it s noted that this prediction is sensitive to the value chosen for the non - perturbative quantity @xmath7 and the ckm matrix element @xmath8 , in this paper , we use their values from refs.@xcite . the implication of @xmath1 measurements have already been studied in model independent approach @xcite , mssm models @xcite , @xmath9-model @xcite , grand unified models @xcite . the sm prediction in eq.([smderict ] ) suffers large uncertainties from the hadronic parameters , nevertheless , the experimental data agree fairly well with the sm value . therefore , we can use the cdf measurement to constrain new physics which may induce the b - s transition . effects of the r - parity violating ( rpv ) supersymmetry ( susy ) on the neutral meson mixing have been discussed extensively in the papers @xcite . in this paper we will consider the rpv susy effects at the tree level in the @xmath2 mixing by the latest experimental data . using the latest experimental data of @xmath1 and the theoretical parameters , we obtain the new bound on the relevant rpv coupling product . if there are rpv contributions to @xmath1 , the same new physics will also contribute to the width difference @xmath10 , and therefore we will use the constrained parameter region to examine the rpv effects on @xmath10 . we first consider the sm contribution to the @xmath2 mixing . the sm effective hamiltonian for the @xmath2 mixing is usually described by @xcite @xmath11^{-6/23 } \left[1+\frac{\alpha_s(\mu_b)}{4\pi}j_5\right]\mathcal{o}+h.c.,\label{hsm}\end{aligned}\ ] ] with @xmath12 where @xmath13 and @xmath14 is the qcd correction . in terms of eq.([hsm ] ) , the mixing amplitude @xmath15 in the sm , dominated by the top quark loop , is @xmath16 defining the renormalization group invariant parameter @xmath17 by @xmath18^{-6/23 } \left[1+\frac{\alpha_s(\mu)}{4\pi}j_5\right],\\ & & \langle b^0_s\|\mathcal{o}\|\bar{b}^0_s\rangle \equiv\frac{8}{3}b_{b_s}(\mu)f^2_{b_s}m_{b_s}^2,\end{aligned}\ ] ] then , we have the @xmath0 mass difference in the sm @xmath19 in the sm , the off - diagonal element of the decay width matrix @xmath20 may be written as @xcite @xmath21,\end{aligned}\ ] ] here @xmath22 , @xmath23 and @xmath24 at the @xmath25 scale @xcite , and the @xmath26 corrections @xmath27 are given in @xcite . the operator @xmath28 can be found in eq.([o ] ) , one now encounters a second operator operator , @xmath29 , and thereby another b - parameter @xmath30 @xmath31 the width difference between @xmath0 mass eigenstates is given by @xmath32,\end{aligned}\ ] ] and the sm predicts @xmath33 with the input parameters in table i @xmath34 it s noted that the width difference have been reviewed recently in @xcite . now we turn to the rpv susy contributions to the @xmath2 mixing . in the most general superpotential of the minimal supersymmetric standard model , the rpv superpotential is given by @xcite @xmath35k}\hat{l}_i\hat{l}_j\hat{e}^c_k+ \lambda'_{ijk}\hat{l}_i\hat{q}_j\hat{d}^c_k+\frac{1}{2 } \lambda''_{i[jk]}\hat{u}^c_i\hat{d}^c_j\hat{d}^c_k , \label{rpv}\end{aligned}\ ] ] where @xmath36 and @xmath37 are the su(2)-doublet lepton and quark superfields , @xmath38 , @xmath39 and @xmath40 are the singlet superfields , while @xmath41 , @xmath42 and @xmath43 are generation indices and @xmath44 denotes a charge conjugate field . mixing . ] the @xmath45 couplings of eq.([rpv ] ) make the @xmath2 mixing possible at the tree level through the exchange of a sneutrino @xmath46 both in the s- and t - channels displayed in fig.[rpvtreelevel ] . the rpv tree level contributions to @xmath2 mixing are described by @xmath47 where we have a new physics operator @xmath48 and we define the b - parameter as @xmath49 note that the expectation values are scaled by factor of @xmath50 over those given in some literature due to our different normalization of the meson wave functions . it is trivial to check that both conventions yield the same values for physical observables . the rpv mixing amplitude @xmath51 is @xmath52 given the expressions above , we now write the total @xmath0 mass difference included both sm and rpv contributions @xmath53 with @xmath54 where the parameters @xmath55 and @xmath56 give the relative magnitude and relative phase of the rpv contribution , i.e. @xmath57 and @xmath58arg@xmath59 . the @xmath0 width difference beyond the sm has been studied in refs.@xcite . if there are rpv contributions to @xmath60 , the same new physics will also contribute to the @xmath0 width difference . the width difference including the rpv contributions is given by @xcite @xmath61 where @xmath62arg@xmath63 , and @xmath64 turns out to be an excellent approximation in the sm . the effect of np on the off - diagonal element of the decay width matrix @xmath65 is anticipated to be negligibly small , hence @xmath66 holds as a good approximation @xcite . we now perform numerical calculation and show the constraint imposed by the measurement of @xmath1 only or both @xmath1 and @xmath67 . the values of the input parameters used in this paper are collected in table i , and we will use the input parameters and the experimental data which vary randomly within @xmath68 variance . [ cols="<,^ " , ] using the constrained parameter space from @xmath1 as shown in fig.[bounds1 ] , one can predict the rpv effects on the @xmath0 width difference @xmath10 . our predictions of @xmath10 are displayed in fig.[dg ] . from fig.[dg](a ) , we find that @xmath69 can have any value from @xmath70 to @xmath71 , as discussed in ref.@xcite , the rpv contributions to the mixing could reduce @xmath67 relative to the sm prediction , and @xmath10 lies between @xmath72 and @xmath73 . we present correlation between @xmath10 and the parameter space of @xmath74 by the three - dimensional scatter plot in fig.[dg](b ) . we also give projections on three vertical planes , where the @xmath75-@xmath76 plane displays the constrained region of @xmath74 as the plot of fig.[bounds1 ] . it s shown that @xmath10 is decreasing first and then increasing with @xmath75 on the @xmath10-@xmath75 plane . from the @xmath10-@xmath76 plane , we can see that @xmath10 may be reduced to zero when @xmath77 lies in @xmath78 $ ] . constrained by the data of @xmath1 and @xmath67 . ] the present experimental data of the @xmath0 width difference have a large error , and we obtain the averaged value from @xcite @xmath79 now we add the experimental constraint of @xmath10 to the allowed space of @xmath74 . we can not get the solution to the experimental data of @xmath10 at @xmath68 level . if @xmath10 is varied randomly within @xmath80 variance , we can obtain the scatter plot as exhibited in fig.[bounds2 ] . comparing fig.[bounds2 ] with fig.[bounds1 ] , we can see that the experimental bound on @xmath10 shown in eq.([dgdata ] ) obviously excludes the region @xmath81 . the stronger limit on @xmath75 from @xmath60 and @xmath67 than the one from @xmath1 only is obtained @xmath82\times 10^{-7}\times \left(\frac{100~gev}{m_{\tilde{\nu}_i}}\right)^2.\end{aligned}\ ] ] in summary , we have studied the rpv tree level effects in the @xmath2 mixing with the current experimental measurements . as shown , using the latest experimental data of @xmath1 and the theoretical parameters , we have obtained the allowed space of the rpv coupling product @xmath74 , the upper bound on the magnitude of @xmath74 has been greatly improved over the existing bounds obtained from the rpv susy . then , we have examined the rpv effects on @xmath67 by the constrained region of @xmath74 from @xmath1 , and we have found that the rpv contributions to the mixing could reduce @xmath67 relative to the sm prediction . finally , using the experimental data of @xmath1 and @xmath67 , we have obtained stronger bound than the one from @xmath1 only on @xmath74 . in addition , we stress that once lhc is turned on , with the anticipated production of @xmath83 @xmath84 per year , the measurements of @xmath1 and @xmath85 will be much more accurate , then the allowed parameter space for @xmath74 will be significantly shrunken or ruled out .
the properties of transport through small conducting islands have been investigated extensively during the past years . electric current in such devices flows due to tunneling of single electrons and is subject to the so - called coulomb blockade effect @xcite which is characterized by a new energy scale , the capacitive charging energy @xmath2 of the island ( see below ) . in the recent past , substantial attention has been devoted also to thermoelectric effects in single - electron devices @xcite . single - electron devices are interesting candidates for thermoelectric applications as it has long been known that dimensional reduction of the electron dynamics may lead to an enhanced thermoelectric efficiency @xcite . while an immense amount of work has been done to investigate thermopower for quantum dots , surprisingly little is known about the thermoelectric effects in single - electron transistors ( set ) with superconducting electrodes . in particular , set with superconducting islands are interesting as they may exhibit the _ parity effect _ where a single unpaired quasiparticle determines the macroscopic thermodynamic properties of the island electrode @xcite as well as the current - voltage characteristics of set with normal - conducting electrodes and a superconducting island ( nsn set ) @xcite . the parity effect can be observed below a crossover temperature @xmath3 for typical system parameters ( here @xmath0 is the energy gap of the superconductor ) . one may expect that the peculiar combination of properties like several competing energy scales ( @xmath2 , @xmath0 , @xmath4 ) and the presence of a singularity in the quasiparticle spectrum gives rise to interesting behavior in the thermoelectric response of such systems . in ref . @xcite the thermopower for an nsn set was studied for @xmath5 and @xmath6 . even for this regime without parity effect , interesting oscillations of the thermopower as a function of the electrostatic island potential and their strong dependence on the ratio @xmath7 were predicted . in this article , we investigate the thermopower of nsn set for temperatures below the crossover temperature . we find that the interplay of energy scales , coulomb - blockade and parity effects , and the peculiarity of the electronic spectrum lead to a rich variety of features in the thermopower @xmath8 . most intriguingly , however , for certain gate voltages this system displays a dramatic enhancement of the thermoelectric efficiency quantified by the figure of merit @xmath1 . this paper is organized as follows . first we introduce the setup and the theoretical methods that are used to describe transport in such systems . subsequently we briefly review the parity effect in set with superconducting islands . we then turn to describe the results for the thermopower , and to interpret them in terms of average transport energies . finally we discuss the surprising results for the figure of merit @xmath1 which we explain in the frame of a simple re - interpretation of this quantity . in an nsn set , a superconducting island with a small electrostatic capacitance @xmath9 is connected via tunnel junctions to two normal - conducting leads ( cf . fig . 1 ) . the corresponding charging energy @xmath10 is large compared to the temperatures under consideration ( here , @xmath11 denotes the elementary charge ) . the conductances of the tunnel junctions are assumed to be small compared to @xmath12 , so sequential tunneling dominates and cotunneling effects may be neglected . the electrostatic potential of the island can be controlled by means of an external potential @xmath13 due to the gate voltage @xmath14 , and the electrostatic energy of the setup with @xmath15 excess electrons on the island can be expressed as @xmath16 the energy cost for adding a single electron to the island is @xmath17 while to add two electrons , an energy @xmath18 is required . a current may flow in the device if a bias voltage @xmath19 and/or a temperature difference @xmath20 is applied . throughout this work we consider the linear - response regime , that is , @xmath21 and @xmath22 . linear transport of charge and heat is conveniently described in terms of the equations @xmath23 which relate the charge and heat current response , @xmath24 and @xmath25 , to an applied potential and temperature difference . here , @xmath26 is the linear ( charge ) conductance , and the thermal conductance @xmath27 is defined via @xmath28 for @xmath29 , _ i.e. _ , @xmath30 . the relevant quantity for the thermoelectric response is the thermopower @xmath8 , given by @xmath31 . in order to calculate the the conductances @xmath26 , @xmath32 and the thermopower @xmath8 we employ a master - equation formalism . according to the orthodox theory @xcite charge and heat current through the system can be written as @xmath33 \\ \label{eq : qcurrent } i_q & = & \sum_n\sum_{j=1,2 } p_n \left [ q_r^{\ , n\to n - j } - q_r^{\ , n\to n+j } \right]\end{aligned}\ ] ] where @xmath34 is the stationary probability for finding @xmath15 excess electrons on the island , @xmath35 is the tunneling rate of @xmath36 electrons from the island to the right lead , and @xmath37 denotes the tunneling rate of @xmath36 electrons from the right lead to the island . correspondingly , @xmath38 is the energy transfer rate in @xmath36-electron tunneling to the right lead whereas @xmath38 denotes the energy transfer rate in a @xmath36-electron tunneling event from the right lead to the island . we consider only sequential tunneling and neglect co - tunneling events . in eqs . , we have taken into account the possibility of single - electron tunneling ( @xmath39 ) and coherent two - electron tunneling ( @xmath40 ) @xcite . the rates for the latter process are given by @xmath41 where @xmath42 , @xmath43 are the andreev conductances in the left and the right junction and @xmath44 , @xmath45 are the energies which are dissipated in a two - electron transfer , _ e.g. _ , @xmath46 . the energy transfer rate for two - electron tunneling is then obtained as @xmath47 where the reference point is the fermi level of the leads . the rates for single - electron transitions are given by a sum of two contributions . on the one hand , we have the standard expressions for tunneling , _ e.g. _ , from a normal to a superconductor @xcite @xmath48\ , \ i = l,\ r\ ] ] ( where @xmath49 denotes the fermi function with the appropriate temperature @xmath50 , @xmath51 , or @xmath52 for the leads or the island , respectively ) . on the other hand , there is the _ escape _ rate of a single unpaired quasiparticle whose energy equals that of @xmath0 @xcite @xmath53 with the normal - electron density of states per spin on the island @xmath54 . note that there is also a corresponding recombination rate . in many cases these escape rates can be neglected , however , at very low temperatures they may exceed the subgap tunneling rate originating from thermally excited quasiparticles in the superconductor . in that case , they produces a different macroscopic behavior of the system , depending on whether the total charge number on the island is even or odd . thus we have @xmath55 where , _ e.g. _ , @xmath56 . finally , the corresponding energy transfer rates for single - electron tunneling are found from @xmath57\ ] ] and @xmath58 ( where again the reference point is the lead fermi level ) such that we have @xmath59 in order to calculate the currents , we solve the stationary master equation for the probabilities @xmath34 @xmath60 for an applied ( small ) bias voltage @xmath61 or a ( small ) temperature difference @xmath62 . if a voltage or temperature difference is applied the island is , strictly speaking , in a non - equilibrium state . for the given physical situation it is reasonable to neglect the non - equilibrium part of the distribution function ( which we have already done by writing the transfer rates in the form above ) . for the temperature @xmath52 of the island we assume the arithmetic mean @xmath63 . we note that there is an independent test of the numerical calculation by checking the onsager relation @xmath64 for the coefficients in eq . which is obeyed to a high accuracy by our method . as mentioned above , the parity effect arises as a macroscopic manifestation of the parity of the electron number in a superconductor , _ i.e. _ , different behavior depending on whether the total electron number is even or odd . the effect was predicted in ref . @xcite and first observed by tinkham _ et . @xcite . in an even - number superconductor at @xmath65 all electrons near the fermi level are bound in cooper pairs and there is not a single unpaired quasiparticle left . this system has an energy gap @xmath66 . on the other hand , adding _ one _ electron results in a single quasiparticle excitation which which does not have an excitation gap . however , the response to an external field is drastically reduced as it is caused just by a single electron . it is intuitively clear that the difference between the two parities will fade away as soon as there are more thermal quasiparticles . this defines the criterion for a crossover temperature @xmath4 beyond which even - odd differences disappear for an isolated superconductor such as the island in the nsn set : it is the temperature at which there is on average one thermally excite quasiparticle and it is defined by @xmath67 where @xmath68 can be viewed as an effective number of accessible quasiparticle states at the temperature @xmath52 . superconducting islands in set are made of aluminum , and for typical parameters ( cf . @xcite ) one has @xmath69 and @xmath70mk . keeping in mind that an set has the characteristic energy scale @xmath2 , there arise four interesting transport regimes : for the gap and the charging energy we may have @xmath5 or @xmath71 . these two cases may be studied for @xmath72 ( no parity effects ) or in the parity regime @xmath73 . interestingly , with our numerical method we have a choice for studying the system with parity effects . for computations at high temperatures @xmath72 it does not matter whether or not the two - electron and escape rates are included , they do not give any observable effect . on the other hand , for @xmath74 even - odd differences can not be observed without including these rates . therefore , including or not including these rates in calculations below @xmath75 helps us to identify the contribution due to the parity - generating processes . before we discuss our results we briefly recall an idea due to matveev @xcite ( cf . also @xcite ) to interpret the thermopower as an average energy @xmath76 at which current is transported in the voltage - biased system ( under linear transport conditions ) . this is an intuitive and powerful method which will use throughout our work to interpret our results . the idea can be understood by taking into account that the current @xmath24 through a device can be written as @xmath77w(\xi){\mathrm d}\xi$ ] where @xmath78 and @xmath79 denote the fermi functions in the left and right lead for the respective potential and temperature ( the energies @xmath80 are taken with respect to the chemical potentials ) . here @xmath81 includes all other quantities such as density of states in the leads and transparency of the device at the energy @xmath80 . by noting that @xmath82 and @xmath83 one finds @xmath84 and consequently for the thermopower @xmath85 that is , up to a factor the thermopower measures directly this average energy @xmath76 . let us turn now to our calculations for the thermopower of nsn set . for the numerics we measure energy and inverse time in units of @xmath2 , and conductance in units of @xmath86 . then , the free parameters are the ratios of the conductances @xmath87 , @xmath88 , @xmath42 , @xmath43 , and the escape rates @xmath89 , @xmath90 . the realistic parameter values which enter our calculations are : @xmath91 , @xmath92 , @xmath93ev , @xmath94 . the experiments in ref . @xcite have been carried out with device parameters close to these values . in fig . 2 we show the results for calculations of the thermopower with and without parity effects . the functional dependence @xmath95 in the latter case has been explained analytically in ref . @xcite . as we have @xmath96 it is sufficient to include two ( or at most three ) charge states in the considerations in order to understand the behavior of @xmath95 . here we focus on a brief discussion of the additional features in the parity regime @xmath74 . = 0.43 [ fig2 ] the key to understand the functional dependence of @xmath95 without even - odd effects is that at the points @xmath97 , that is , when @xmath98 takes half - integer values tunneling to the island @xmath99 and from the island @xmath100 occurs with equal probability . by using simple arguments for the probabilities @xmath34 and @xmath101 the following equation has been derived in ref . @xcite @xmath102 \right)\ ] ] which governs the behavior of @xmath95 in the intervals from @xmath103 to the zeros of @xmath95 which are closest to those @xmath98 values . by inspecting fig . 2 we note that the essential difference introduced in the parity regime is that the purple curve is shifted from the half - integer values of @xmath98 towards the closest even number . as the slope of the curve does not change in this shift , the consequence is that the maximum absolute value of @xmath8 increases , compared to the case without parity . it is not difficult to describe this behavior analytically by repeating the arguments that lead to eq . , now taking into account that the recombination rate of quasiparticles on the island is no longer @xmath104 . in this range of @xmath98 there is an unpaired quasiparticle on the island whose recombination rate provides the dominating contribution to the charge current . this rate is @xmath105 . from this we get a modified relation @xmath106 \right)\ ] ] with @xmath107 } $ ] . the relation captures the essential features of the purple curve in fig . 2 . the green dashed line in fig . 3 shows an example for the thermopower without parity effects for @xmath71 . the arguments we have given in the preceding subsection for the thermopower at half - integer values of @xmath98 are valid also in this case . that is , eq . correctly describes the behavior of the function @xmath95 also in this parameter range . hence , this equation turns out to be the key for understanding the thermopower in nsn set , _ i.e. _ , a device with an electronic spectrum that is gapped around the fermi energy . we further mention that for the thermopower without parity effects for the average energy close to integer values of @xmath108 one has @xmath109 which yields @xmath110 = 0.43 [ fig3 ] if the gap exceeds the charging energy , two - electron tunneling starts to play a prominent role for the current - voltage characteristics of nsn set in the parity regime @xmath74 @xcite . around odd - integer values of @xmath98 there occurs a current peak which is due to a cycle of two - electron tunneling processes . note that these processes do not have a gap . the thermopower is analogous to that of single - electron tunneling in nnn set , the only difference is that we have to substitute the charging energy difference for two - electron tunneling , @xmath111 which leads to a thermopower @xmath112 . this result curiously coincides with the one in the absence of parity effects , eq . although the dominating tunneling process is a different one . in the vicinity of the half - integer values of @xmath98 we observe the analogous effect to the case @xmath5 : the curve gets shifted towards the closest even - integer @xmath98 without changing the slope . scrutiny of the dominating current - carrying processes , _ e.g. _ , for @xmath113 reveals that also here the largest rate is due to quasiparticle recombination on the island . however , if @xmath74 there are essentially no thermal quasiparticles . there is only a _ single _ unpaired electron which is left from the pair - breaking tunneling off the island . the quantitative description of this recombination process leads in full analogy to the conclusion of the previous subsection , namely that eq . describes this shift of the curve . the essence also here is that the current is due to tunneling of a single unpaired quasiparticle . once we have studied the thermopower of the nsn set it is an interesting question to investigate the thermoelectric efficiency of this device . this efficiency is quantified by @xmath114 where @xmath115 and @xmath116 denote the electronic and lattice contribution to the heat conductance of the device . our work focuses on a regime of extremely low temperatures for which @xmath116 is very small , hence we neglect it and @xmath117 . we compute @xmath27 from the heat current @xmath25 according to eqs . . an example of the results is shown in fig . 4 . for @xmath118 we observe huge values of @xmath1 which is rather uncommon , keeping in mind that typical = 0.43 [ fig4 ] values for materials reach the order of 1 , or for quantum dots the order of @xmath119 . the question is why such high values are possible for this system . a quick answer might be that , given that superconductors are bad heat conductors , it could be expected that by including the superconducting island @xmath1 of a single - electron transistor would be enhanced . let us try to give a more quantitative answer which takes into account our observations from the previous section . by transferring matveev s idea for the interpretation of @xmath8 as an average transport energy of the electrons according to eq . we can derive an expression that illuminates the meaning of @xmath1 . in analogy with eq . we obtain @xmath120 . this leads , together with eq . and @xmath30 , to the new relation @xmath121 indeed , all additional factors cancel and @xmath1 turns out to be the ratio of the squared average transport energy and the variance of that energy . this relation clearly indicates the strategy that needs to be used in order to increase @xmath1 : the current - carrying electrons should be far from the fermi energy while their energetic distribution should be as narrow as possible . this corroborates also the conclusion of ref . @xcite that a @xmath122 function in the spectrum is favorable for a high @xmath1 value . let us now apply eq . to the nsn set . we have already discussed that for the @xmath98 values where the strong enhancement of @xmath1 is found , the current is carried by a single unpaired quasiparticle . the maximum energy of this quasiparticle is @xmath123 while its energy distribution is rather narrow : it is just given by the temperature @xmath52 . hence we expect @xmath124 . in fact , this estimate has the correct order of magnitude . we have investigated the thermopower for a single - electron transistor with a superconducting island for arbitrary ratios @xmath7 and for temperatures both above and below the crossover temperature for parity effects @xmath75 . the results show the expected parity effects also in the functional dependence @xmath95 of the thermopower on the gate charge @xmath98 , in particular @xmath125 periodicity . we have provided a discussion of the essential features of this functional dependence in terms of eqs . , . it is remarkable that the basis for this discussion is eq . which was found already in ref . @xcite . apart from the thermopower we have also calculated the thermoelectric figure of merit @xmath1 . unexpectedly we have found a strong enhancement of @xmath1 compared to common values of this quantity . in order to understand our findings we have given a new interpretation of @xmath1 in terms of matveev s idea to represent thermoelectric quantities as moments of the energy distribution for the current - carrying electrons . it shows that large values of @xmath1 can be obtained if the dominant transport mechanism occurs far from the fermi level , and at the same time , has a narrow distribution in its energies . clearly , the nsn set is a system where these conditions can be achieved . financial support through priority program 1386 of the german research foundation is gratefully acknowledged . averin and k.k . likharev , in _ mesoscopic phenomena in solids _ , edited by b. altshuler , p.a . lee , and r.a . webb , elsevier amsterdam ( 1991 ) . beenakker and a.a.m . staring , phys . b * 46 * , 9667 ( 1992 ) a.a.m . staring et al , europhys . lett . * 22 * , 57 ( 1993 ) a.s . dzurak et al , phys . b * 55 * , r10197 ( 1997 ) m. turek and k.a . matveev , phys . b * 65 * , 115332 ( 2002 ) s. mller , h. buhmann , s.f . godijn , and l.w . molenkamp , phys . lett . * 81 * , 5197 ( 1998 ) a.v . andreev and k.a . matveev , phys . lett . * 86 * , 280 ( 2001 ) k.a . matveev and a.v . andreev , phys . b * 66 * , 045301 ( 2002 ) d. boese and r. fazio , europhys . lett . * 56 * , 576 ( 2001 ) j. koch , f. von oppen , y. oreg , and e. sela , phys . rev . b * 70 * , 195107 ( 2004 ) m. turek , j. siewert , and k. richter , phys . rev . b * 71 * , 220503(r ) ( 2005 ) . b. kubala , j. knig , and j. pekola , phys . lett . * 100 * , 066801 ( 2008 ) . r. scheibner , m. knig , d. reuter , a.d . wieck , c. gould , h. buhmann , and l.w . molenkamp , new . j. phys . * 10 * , 083016 ( 2008 ) . kuo , j. appl . phys . * 48 * , 125005 ( 2009 ) . t. costi and v. zlatic , phys . b * 81 * , 235127 ( 2010 ) . m. tsaousidou and g.p . triberis , j. phys . : condens . matter * 22 * , 355304 ( 2010 ) . m. wierzbicki and r. swirkowicz , phys . b * 84 * , 075410 ( 2011 ) . hicks and m.s . dresselhaus , phys . b * 47 * , 12727 ( 1993 ) . mahan and j.o . sofo , proc . sci . * 93 * , 7436 ( 1996 ) . averin and yu.v . nazarov , phys . * 69 * , 1993 ( 1992 ) . tuominen , j.m . hergenrother , t.s . tighe , and m. tinkham , phys . lett . * 69 * , 1997 ( 1992 ) . eiles , j.m . martinis , and m.h . devoret , phys . * 70 * , 1862 ( 1993 ) . hergenrother , m.t . tuominen , and m. tinkham , phys . lett . * 72 * , 1742 ( 1994 ) . g. schn , j. siewert , and a.d . zaikin , physica b * 203 * , 340 ( 1994 ) . hekking , l.i . glazman , k.a . matveev , and r.i . shekhter , phys . lett . * 70 * , 4138 ( 1993 ) . abrikosov , _ fundamentals of the therory of metals _ ( north - holland , amsterdam , 1988 ) k.a . matveev , statistical and dynamical aspects of mesoscopic systems . proceedings of the xvi sitges conference on statistical mechanics ( 2000 ) g. schn and a.d . zaikin , europhys . * 26 * , 695 ( 1994 ) .	it is well - known that the transport properties of single - electron transistors with a superconducting island and normal - conducting leads ( nsn set ) may depend on whether or not there is a single quasiparticle on the island . this parity effect has pronounced consequences for the linear transport properties . here we analyze the thermopower of nsn set with and without parity effect , for entirely realistic values of device parameters . besides a marked dependence of the thermopower on the superconducting gap @xmath0 we observe an enhancement in the parity regime which is accompanied by a dramatic increase of the thermoelectric figure of merit @xmath1 . the latter can be explained within a simple re - interpretation of @xmath1 in terms of averages and variances of transport energies .
brownian motion is a stochastic process naturally associated to any riemannian manifold . the distance between brownian motion and a submanifold was studied by the author in @xcite , using an inequality for the laplacian of the distance function based on jacobi field comparison . the brownian bridge is given by conditioning brownian motion to hit a fixed point at a fixed positive time . we extend this concept by replacing the fixed point with a submanifold . there has so far been very little research on this topic , even if the ambient space is euclidean , although such processes have appeared in the context of wiener measure approximation , as in @xcite . suppose that @xmath0 is a complete and connected riemannian manifold of dimension @xmath1 and that @xmath2 is a closed embedded submanifold of @xmath0 of dimension @xmath3 . assume for this introduction that @xmath0 is compact . denote by @xmath4 the heat kernel on @xmath0 and define @xmath5 where @xmath6 denotes the measure on @xmath2 induced by the embedding . for the case in which @xmath2 is a point , @xmath6 is a delta measure and the integral reduces to a pointwise evaluation . the integrated heat kernel will be the central object of study . some examples of it are given in section [ sec : localtime ] . following @xcite , it can be identified as the time derivative of a mean hypersurface local time . a comparison theorem for this object is included in subsection [ ss : hltcomp ] . in section [ sec : bbridges ] , denoting by @xmath7 a brownian motion on @xmath0 starting at @xmath8 , we fix @xmath9 and prove that if @xmath10 then for a bounded @xmath11-measurable random variable @xmath12 we have @xmath13 = \frac{\mathbb{e}\left [ p^m_{t - t}(x_t(x),n)f \right]}{p^m_t(x , n)}.\ ] ] this gives rise to a diffusion over the time interval @xmath14 starting at @xmath8 and arriving in @xmath2 at time @xmath15 , with time - dependent infinitesimal generator @xmath16 we call this a _ brownian bridge to a submanifold_. to show that it is a semimartingale on @xmath17 $ ] we prove the gradient estimate @xmath18 and derive a hessian estimate as corollary . these estimates are given in section [ sec : deriv ] by theorem [ th : derivativeestiamte ] and corollary [ cor : hesscor ] , respectively . we prove them using the method of stroock @xcite , the inductive element of which had previously been discovered by cheng , li and yau @xcite . our estimates generalize the main theorem of engoulatov @xcite and the gradient and hessian estimates of hsu @xcite , who considered only the one point case . to prove the gradient estimate we require a suitable lower bound on @xmath19 . in section [ sec : fbridges ] we define @xmath20 and introduce the diffusion on @xmath0 starting at @xmath8 with time dependent generator @xmath21 where @xmath22 denotes differentiation in the radial direction . we call such a process a _ fermi bridge_. it is an example of a _ submanifold bridge process _ , by which we mean a brownian motion with drift which arrives in a submanifold at a fixed positive time . this terminology was introduced by ndumu @xcite . in section [ sec : sechkform ] we prove the main result of this article , theorem [ th : maintheorem ] , which in terms of a fermi bridge @xmath23 yields the formula @xmath24 \lim_{t\uparrow t}\mathbb{e}\left[\exp\left[\int_0^t \frac{r_n(\hat{x}_s(x))}{t - s } \bigg(d\mathbf{a}_s+d\mathbf{l}_s\bigg)\right]\right].\\[3.0mm]\ ] ] here @xmath25 is an absolutely continuous random measure , which takes into account the geometry of @xmath0 in between @xmath2 and the cut locus , while @xmath26 is a singular continuous random measure , which takes in to account the effect of the cut locus itself . if @xmath2 is a point , this yields a formula for the heat kernel itself . theorem [ th : maintheorem ] is proved using a modification of the method of elworthy and truman @xcite . elworthy and truman s heat kernel formula and ndumu s extension of it to integrated heat kernels @xcite typically require some invertibility of the exponential map and so can not be applied directly to manifolds with a cut locus . the problem of extending these formulae to manifolds with a cut locus was posed to the author by xue - mei li . theorem [ th : maintheorem ] provides this extension and can also be adapted to include a drift and a potential , in analogy with the schrdinger equation for a magnetic field , considered by watling @xcite . the formulae in @xcite , @xcite and @xcite are all expressed in terms of a stochastic process called the _ semiclassical bridge_. the semiclassical bridge will be used to obtain derivative formulae for kernels of schrdinger operators in a future publication by the author and xue - mei li . it is similar to the fermi bridge but has an additional drift which depends on the jacobian determinant of the exponential map . this additional drift results in the appearance of a potential function which is difficult to understand geometrically . the random measure @xmath25 can , on the other hand , be written explicitly in terms of jacobi fields . this allows us in section [ sec : lbar ] to use theorem [ th : maintheorem ] and the estimate on @xmath27 given in @xcite to deduce a lower bound on the integrated heat kernel . this bound is stated in theorem [ th : quadcurvlb ] . it is used to prove an asymptotic relation , stated in theorem [ th : genvar ] , and in section [ sec : deriv ] it is used to prove the desired gradient and hessian estimates . potential applications of these results are discussed in subsection [ ss : furtherapps ] . although for this introduction we have assumed that @xmath0 is compact , this is not always necessary . in fact , we will usually only require some control on the laplacian of the distance function . geometric conditions for this were given in @xcite . in particular , it was proved there that if there exists a function @xmath28 such that one of the following conditions is satisfied off the union of @xmath2 and its cut locus : ( c1 ) : : @xmath3 , the sectional curvature of planes containing the radial direction is bounded below by @xmath29 and the absolute value of the principal curvature of @xmath2 is bounded by a constant @xmath30 ; ( c2 ) : : @xmath31 and the ricci curvature in the radial direction is bounded below by @xmath32 ; ( c3 ) : : @xmath33 , the ricci curvature in the radial direction is bounded below by @xmath32 and the absolute value of the mean curvature of @xmath2 is bounded by a constant @xmath34 ; then the inequality @xmath35 holds off the cut locus pointwise and , by @xcite , on the whole of @xmath0 in the sense of distributions . conditions of this type are used to deduce theorem [ th : quadcurvlb ] . the problem which originally motivated this piece of work was to extend elworthy and truman s formula to manifolds with a cut locus . theorem [ th : maintheorem ] was the first new result to be proved in this article and led to the remainder of the investigation . the author wishes to thank xue - mei li for for advising this and the study of point to submanifold bridges , as either conditioned brownian motions or as brownian motions with directed drift , and for suggesting several useful references . suppose that @xmath0 is a complete and connected riemannian manifold of dimension @xmath1 and that @xmath7 is a brownian motion on @xmath0 starting at @xmath36 defined up to an explosion time @xmath37 . an open connected subset @xmath38 of @xmath0 is called a _ regular domain _ if it has smooth boundary and compact closure . if @xmath39 denotes the dirichlet heat kernel on a regular domain @xmath38 then @xmath39 is the fundamental solution to the heat equation on @xmath38 with dirichlet boundary conditions . if @xmath40 is a non - negative measurable function on @xmath38 then @xmath39 satisfies @xmath41 = \int_m f(y)p^d_t(x , y)\,d\operatorname{vol}_m(y)\ ] ] for all @xmath42 where @xmath43 denotes the first exit time of @xmath7 from @xmath38 . if @xmath44 is an exhaustion of @xmath0 by regular domains then the minimal heat kernel on @xmath0 is given by @xmath45 . it is the minimal fundamental solution of the heat equation on @xmath0 and coincides with the transition densities of brownian motion . on @xmath46 the heat kernel is given by the gauss - weierstrass kernel @xmath47 for @xmath48 and @xmath42 . on @xmath49 , the hyperbolic space of dimension @xmath50 with constant sectional curvatures @xmath51 , there is the formula @xmath52\underbrace{\frac{\sqrt{-{\kappa}}d(x , y)e^{\frac{{\kappa}t}{2}}}{\sinh \left(\sqrt{-{\kappa}}d(x , y)\right)}}_{\leq 1}\ ] ] for @xmath53 and @xmath42 . on @xmath54 , the unit circle , there is the summation formula @xmath55\underbrace{\sum_{k \in \mathbb{z } } \exp\left[-\frac{2\pi k ( d(x , y)+\pi k)}{t}\right]}_{\geq 1}\ ] ] for @xmath56 and @xmath42 , a result of the fact that @xmath54 is covered by @xmath57 . the articles @xcite and @xcite provide iterative formulae for the heat kernels on the standard hyperbolic and spherical spaces , respectively , of arbitrary dimension . it is generally difficult to find deterministic formulae , unless the manifold in question exhibits a certain amount of symmetry , as in the above examples . now suppose that @xmath2 is a closed embedded submanifold of @xmath0 of dimension @xmath3 and recall the definition of the integrated heat kernel given by formula . if @xmath38 is a regular domain in @xmath0 then define similarly @xmath58 for @xmath36 and @xmath42 , where @xmath6 denotes the measure on @xmath2 induced by the embedding . for @xmath59 denote by @xmath60 the circle of radius @xmath61 centred at the origin in @xmath62 . then @xmath63\operatorname{besseli}\left(0,\frac{r\\|x\\|}{t}\right)\ ] ] for @xmath64 and @xmath42 , where @xmath65 denotes the modified bessel function of the first kind . a plot of the integrated kernel is given below in figure [ fig : ring ] . , at a fixed small time @xmath42 . the origin is located at the center of the image.,scaledwidth=40.0% ] for @xmath66 one can think of @xmath67 as a solution to the heat equation on @xmath0 for the measure - valued initial condition given by the dirac measure based at @xmath68 . similarly , @xmath69 can be thought of as a solution to the heat equation on @xmath0 for the measure - valued initial condition @xmath6 . for example , if @xmath2 is a closed embedded surface in @xmath70 uniformly heated at time zero then @xmath71 describes how the heat diffuses for positive times . alternatively , if @xmath2 is a closed embedded loop in @xmath70 then @xmath71 could be used to describe the temperature around a hot metal wire . a probabilistic interpretation of the integrated heat kernel will be given in the next section by theorem [ th : bridgecons ] . our basic approach will be to compare @xmath69 to the function @xmath72\ ] ] since if @xmath73 is viewed as an affine linear subspace of @xmath46 then @xmath74 viewing the @xmath75-dimensional hyperbolic space @xmath76 as an embedded totally geodesic submanifold of @xmath77 , ndumu @xcite used formula with @xmath78 to show that @xmath79 for @xmath80 and @xmath42 . suppose in this section that @xmath0 is a stochastically complete and that @xmath2 is compact . fix @xmath9 and @xmath81 and consider the associated canonical probability space @xmath82 equipped with canonical filtration @xmath83 . here @xmath84 is the space of continuous paths defined on @xmath17 $ ] taking values in @xmath0 , @xmath85 is the @xmath86-algebra generated by the coordinate maps upto time @xmath87 and @xmath88 is wiener measure , with respect to which the coordinate process @xmath89\rbrace$ ] is a brownian motion on @xmath0 starting at @xmath8 . recall that the _ normal exponential map _ is simply the exponential map restricted to the normal bundle @xmath90 . denote its jacobian determinant by @xmath91 . inequalities for this object were provided in @xcite . the following theorem shows how the integrated heat kernel appears naturally when a brownian motion is conditioned to arrive in @xmath2 at the fixed time @xmath15 . [ th : bridgecons ] choose @xmath10 and a bounded @xmath85-measurable random variable @xmath12 . then @xmath92 = \frac { \mathbb{e}^{x}\left [ p^m_{t - t}(x_t , n)f(x ) \right]}{p^m_t(x , n)}.\ ] ] for sufficiently small @xmath93 it follows from the definition of conditional expectation , the markov property , fubini s theorem and the smooth coarea formula that @xmath94\\ = & \text { } \frac{\mathbb{e}^{x}\left[\mathbf{1}_{\lbrace d(x_t , n ) < \epsilon\rbrace}f(x)\right]}{\mathbb{p}^{x}\lbrace d(x_t , n ) < \epsilon\rbrace}\\ = & \text { } \frac{\mathbb{e}^{x } \left[p_{t - t}\mathbf{1}_{\lbrace d(\cdot , n ) < \epsilon\rbrace}(x_t ) f(x ) \right]}{\mathbb{p}^{x}\lbrace d(x_t , n ) < \epsilon\rbrace}\\ = & \text { } \frac{\int_{{b}_{\epsilon}(n)}\mathbb{e}^{x}\left [ p^m_{t - t}(x_t , y)f(x ) \right]\,d\operatorname{vol}_m(y)}{\int_{{b}_{\epsilon}(n ) } p^m_t(x , y)\,d\operatorname{vol}_m(y)}\\ = & \text { } \frac{\int_n \int_{{b}_\epsilon^z(0)}\mathbb{e}^{x}\left [ p^m_{t - t}(x_t,\operatorname{exp}(\xi))f(x ) \right]\theta_n(\xi)\,d\xi\ , d\operatorname{vol}_n(z)}{\int_n \int_{{b}_\epsilon^z(0 ) } p^m_t(x,\operatorname{exp}(\xi))\theta_n(\xi)\,d\xi\ , d\operatorname{vol}_n(z ) } \end{split}\ ] ] where @xmath95 denotes the open ball in @xmath96 of radius @xmath97 centred at the origin . since the volume of these balls is constant and independent of @xmath98 , it follows by the continuity of the above integrands and @xmath99 that @xmath100=\frac{\int_n \mathbb{e}^{x}\left [ p^m_{t - t}(x_t , z)f(x ) \right]\,d\operatorname{vol}_n(z)}{\int_n p^m_t(x , z)\,d\operatorname{vol}_n(z)}\ ] ] from which the result follows , by the definition of the left - hand side of as a radon - nikodym derivative . for each @xmath10 it follows , by theorem [ th : bridgecons ] and the smooth coarea formula , that conditioning a brownian motion to be in the _ interior _ of a tubular neighbourhood of @xmath2 of radius @xmath97 at time @xmath15 while separately conditioning brownian motion to belong to the _ boundary _ of that tubular neighbourhood at time @xmath15 results in two measures on @xmath101 which converge weakly to the same limit as @xmath102 . if for @xmath66 we define a measure @xmath103 on @xmath104 by @xmath105 , for @xmath106 , then theorem [ th : bridgecons ] implies that @xmath103 is absolutely continuous with respect to @xmath88 on @xmath101 for any @xmath107 and that the radon - nikodym derivative is given by @xmath108 in particular , if @xmath103 exists as a probability measure on the space of continuous paths starting at @xmath8 and terminating at @xmath68 at time @xmath15 then under @xmath103 and for @xmath109 the joint density function of @xmath110 , denoted by @xmath111 , is given by @xmath112 as is well - known . we will prove the existence of @xmath103 assuming that there exist constants @xmath113 such that @xmath114\ ] ] for all @xmath115 and @xmath116 $ ] and a constant @xmath117 such that @xmath118\ ] ] for all @xmath119 and @xmath120 . by @xcite and @xcite , such bounds exist if the ricci curvature is bounded below by a constant with @xmath121 lower regular . in particular , by @xcite and @xcite , such bounds exist if the ricci curvature is bounded with @xmath0 having positive injectivity radius , in which case the constants @xmath122 and @xmath123 can be chosen so that @xmath123 is arbitrarily close to @xmath75 . for the case in which @xmath0 is compact , a version of the following lemma was proved by driver in @xcite . [ lem : concrit ] assume and , suppose @xmath124 and without loss of generality assume @xmath125 . then for all @xmath126 there exists a constant @xmath127 such that for all @xmath128 we have @xmath129 \leq \frac{c({m},c,\sigma^2,\beta,\gamma , t)}{p^m_t(x , y)}\frac{\gamma\left(\frac{{m}+{p}}{2}\right)}{\gamma\left(\frac{{m}}{2}\right)}\left(\frac{\sigma^2(t - s)}{1-\gamma}\right)^{\frac{{p}}{2}}\ ] ] so long as @xmath130 . first assume @xmath131 . if @xmath132 then @xmath133r^{{m}-1}ds = \frac{\gamma({m}/2)}{2a^{\frac{{m}}{2}}}\ ] ] and so for @xmath119 we have @xmath134d^{p}(w , z)\,d\operatorname{vol}_m(z)\\ \nonumber \leq & \text { } c e^\beta ( t - s)^{-\frac{{m}}{2}}\int_{t_wm } \\| \xi\\|^{{p}}\exp\left[\left(\beta-\frac{1}{\sigma^2(t - s)}\right)\\| \xi\\|^2\right]dv\\ \nonumber = & \text { } c e^\beta \frac{{m}\pi^{\frac{{m}}{2}}(t - s)^{-\frac{{m}}{2}}}{\gamma\left(\frac{{m}}{2}+1\right)}\int_0^\infty r^{{p}+{m}-1}\exp\left[\left(\beta-\frac{1}{\sigma^2(t - s)}\right)r^2\right]dr\\ \nonumber = & \text { } c e^\beta\frac { \pi^{\frac{{m}}{2}}\gamma\left(\frac{{m}+{p}}{2}\right)(t - s)^{-\frac{{m}}{2}}}{\gamma\left(\frac{{m}}{2}\right)}\left(\frac{1-\beta\sigma^2(t - s)}{\sigma^2(t - s)}\right)^{-\frac{({m}+{p})}{2}}\\ \nonumber \leq & \text { } c e^\beta\frac{\gamma\left(\frac{{m}+{p}}{2}\right)}{\gamma\left(\frac{{m}}{2}\right)}\left(\frac{\sigma^2\pi}{1-\gamma}\right)^{\frac{{m}}{2}}\left(\frac{\sigma^2(t - s)}{1-\gamma}\right)^{\frac{{p}}{2}}.\nonumber\end{aligned}\]]thus there exists a constant @xmath135 such that @xmath136 for all @xmath137 , @xmath119 and @xmath138 satisfying @xmath130 . for such @xmath138 we see that @xmath139\\ = & \text { } \int_m \int_m \frac{p^m_s(x , w)p^m_{t - s}(w , z)d^{{p}}(w , z ) p^m_{t - t}(z , y)}{p^m_t(x , y ) } d\operatorname{vol}_m(w)d\operatorname{vol}_m(z ) \\ \leq & \text { } ( t/3)^{-{m}/2}\frac { c_1({m},c,\sigma^2,\beta,\gamma)}{p^m_t(x , y ) } \frac{\gamma\left(\frac{{m}+{p}}{2}\right)}{\gamma\left(\frac{{m}}{2}\right)}\left(\frac{\sigma^2(t - s)}{1-\gamma}\right)^{\frac{{p}}{2}}. \nonumber \end{split}\ ] ] the same result is obtained for @xmath140 . the cases @xmath141 or @xmath142 can be treated similarly . if @xmath12 is a bounded @xmath85-measurable function on @xmath84 for some @xmath10 then it follows from theorem [ th : bridgecons ] that @xmath143 = \frac{\int_n p^m_t(x , y)\mathbb{e}^{x , y;t}\left [ f(x ) \right]\,d\operatorname{vol}_n(y)}{p^m_t(x , n)}\ ] ] which implies , by lemma [ lem : concrit ] , that for all @xmath144 there exists @xmath145 such that @xmath146 \leq c_\epsilon(t - s)^{\frac{{p}}{2}}\nonumber\ ] ] for all @xmath124 with @xmath147 . it follows from kolmogorov s continuity theorem , by covering the interval @xmath17 $ ] with finitely many closed intervals each of length less that @xmath97 , that there exists a probability measure @xmath148 on the bridge space @xmath149 which satisfies @xmath150 for @xmath151 and whose finite - dimensional distributions can be deduced from equations and . in particular , if @xmath152 denotes the law of the random variable @xmath153 under the measure @xmath148 then @xmath154 for example , if @xmath155 with @xmath2 the unit @xmath156-sphere and @xmath157 then @xmath152 is given by the uniform measure on @xmath2 . in contrast , if @xmath155 with @xmath2 an @xmath158-dimensional subspace and @xmath157 then @xmath152 is given by the heat kernel measure on @xmath2 . note that the measure @xmath148 clearly exists in the latter case , even though @xmath2 is non - compact . [ lem : concrit ] we can prove the following concentration inequality for tubular neighbourhoods . [ prop : concencondthm ] assuming and , for all @xmath126 there exists @xmath159 such that @xmath160 for all @xmath161 $ ] . without loss of generality , assume @xmath162 . since @xmath163 it follows that for each @xmath164 we have @xmath165 \leq \mathbb{e}^{x , y;t}\left[d^p(x_t , y)\right]$ ] . therefore , by applying with @xmath166 , we see by lemma [ lem : concrit ] that for all @xmath126 there exists @xmath167 such that @xmath168\leq \frac{c({m},c,\sigma^2,\beta,\gamma , n , t)}{p^m_t(x , n)}\frac{\gamma\left(\frac{{m}+{p}}{2}\right)}{\gamma\left(\frac{{m}}{2}\right)}\left(\frac{\sigma^2(t - t)}{1-\gamma}\right)^{\frac{{p}}{2}}\nonumber\ ] ] for all @xmath169 with @xmath170 . for such @xmath87 , choosing @xmath171 , applying this bound to the case where @xmath172 is an even integer and summing yields @xmath173 \leq c({m},c,\sigma^2,\beta,\gamma , x , n , t ) \left(1- \frac{\theta\sigma^2(t - t)}{2(1-\gamma)}\right)^{-\frac{{m}}{2 } } \nonumber\ ] ] so long as @xmath174 . under these conditions on @xmath87 , it follows from markov s inequality that for all @xmath59 there is the estimate @xmath175 fixing @xmath176 and choosing @xmath177 this yields @xmath178 from which the result follows since @xmath179 can be chosen arbitrarily close to @xmath180 . it follows from and girsanov s theorem that under the measure @xmath148 the coordinate process on @xmath84 is a diffusion on the half - open time interval @xmath14 starting at @xmath8 with time - dependent infinitesimal generator @xmath181 for @xmath182 . to show that it is a semimartingale under this measure over the closed time interval @xmath17 $ ] we require a suitable estimate on the logarithmic derivative . this estimate will be deduced from the results obtained in sections [ sec : fbridges ] , [ sec : sechkform ] and [ sec : lbar ] . it is given by theorem [ th : derivativeestiamte ] in section [ sec : deriv ] and the semimartingale property is given by corollary [ cor : hesscor ] . in this section we introduce a bridge process defined in terms of the distance function @xmath183 . recall that the cut locus of @xmath2 , denoted by @xmath184 , is given by the closure of the set of points at which @xmath185 fails to be differentiable . it is a closed set of @xmath121-measure zero which is , upto a set of hausdorff dimension at most @xmath186 , an at most countable union @xmath187 of open subsets of hypersurfaces . these facts follow from the work of angulo ardoy and guijarro @xcite and mantegazza and mennucci @xcite who used hamilton - jacobi equations and the theory of viscosity solutions . the vector field @xmath22 will denote differentiation in the radial direction , defined off the union of @xmath2 and @xmath184 to be the gradient of @xmath188 and set equal to zero elsewhere . for @xmath9 fixed with @xmath189 defined by equation , the time - dependent vector field @xmath190 where @xmath182 is smooth away from the cut locus but generally not continuous on it . one imagines the deterministic flow associated to this vector field as being one for which @xmath184 is a source and for which @xmath2 is a sink . the strength of the flow increases dramatically as the terminal time @xmath15 is approached , while the vector field vanishes on @xmath2 . a diffusion on @xmath0 starting at @xmath8 with time - dependent infinitesimal generator @xmath21 will be called a _ fermi bridge between @xmath8 and @xmath2 in time @xmath15_. such processes will always be defined upto a predictable stopping time less than or equal to @xmath15 . we use the name _ fermi bridge _ since the time - dependent part of the drift acts in a direction normal to @xmath2 , which would be the radial part of a system of polar fermi coordinates , and since there are conditions under which this process arrives at @xmath2 at time @xmath15 almost surely . note that if @xmath155 with @xmath2 a point then the above definition reduces to that of a standard brownian bridge . suppose that @xmath23 is a fermi bridge between @xmath8 and @xmath2 in time @xmath15 , defined upto the minimum of @xmath15 and its explosion time . suppose also that @xmath38 is a regular domain in @xmath0 and denote by @xmath191 the first exit time of @xmath23 from @xmath38 . since @xmath192 is polar for @xmath23 and since the martingale part of an antidevelopment of @xmath23 is a standard brownian motion , the it formula of barden and le ( * ? ? * thoerem 1 ) implies that there exist continuous adapted non - decreasing and non - negative processes @xmath193 and @xmath194 whose associated random measures are singular with respect to lebesgue measure and supported when @xmath23 takes values in @xmath2 and @xmath187 , respectively , such that @xmath195 & \text { } \quad\quad -\int_0^{t \wedge \hat{\tau}_d } \frac{r_n(\hat{x}_s(x))}{t - s}\,ds-\mathbb{l}^{\operatorname{cut}(n)}_{t \wedge \hat{\tau}_d}(\hat{x}(x ) ) + l^n_{t \wedge \hat{\tau}_d}(\hat{x}(x ) ) \end{split}\ ] ] for all @xmath107 , almost surely , where @xmath196 is a standard one - dimensional brownian motion and @xmath197 here @xmath198 is any unit normal vector field on @xmath187 and the gteaux derivatives @xmath199 are defined for @xmath200 and @xmath201 by @xmath202 and @xmath203 . the processes @xmath204 and @xmath194 are , roughly speaking , given by the local times at zero of @xmath205 and @xmath206 , respectively , whenever the latter make sense . using this formula we can estimate the radial moments of the fermi bridge , restricting our attention to the domain @xmath38 and doing so under the assumption that there exist constants @xmath207 and @xmath208 such that @xmath209 on @xmath210 . [ th : secradmomthmbr ] let @xmath207 and @xmath208 be any constants such that inequality holds on @xmath211 . then we have @xmath212 \leq \left(r_n^2 ( x ) \left ( \frac{t - t}{t}\right ) + \nu t \right ) \left ( \frac{t - t}{t}\right ) e^{\lambda t}\ ] ] for all @xmath107 . define the function @xmath213 by @xmath214 \nonumber\ ] ] for @xmath10 . by it s formula and formula we deduce the differential inequality @xmath215 for all @xmath10 . applying gronwall s inequality to it yields @xmath216 where we used the assumption @xmath208 for the second inequality . note that for @xmath155 with @xmath2 a linear subspace , with @xmath217 and @xmath218 , one can set @xmath219 and inequality holds as an _ equality_. also , jensen s inequality implies the following estimate on the first radial moment . [ cor : frmebr ] let @xmath207 and @xmath208 be any constants such that inequality holds on @xmath211 . then we have @xmath220 \leq \left(\left(r_n^2 ( x ) \left ( \frac{t - t}{t}\right ) + \nu t\right ) \left ( \frac{t - t}{t}\right)\right)^{\frac{1}{2 } } e^{\frac{\lambda t}{2}}\ ] ] for all @xmath107 . in the next subsection we consider the case where there exist constants @xmath207 and @xmath208 such that inequality holds on the whole of @xmath221 . suppose for this subsection that there exists constants @xmath207 and @xmath208 such that inequality holds on @xmath221 . suppose also that @xmath7 is a _ non - explosive _ brownian motion on @xmath0 starting at @xmath8 , defined on a filtered probability space @xmath222 satisfying the usual conditions . note that , by ( * ? ? ? * theorem 5 ) , if @xmath2 is compact then inequality _ implies _ the non - explosion of @xmath7 . for @xmath10 define @xmath223 \nonumber\ ] ] where @xmath224 is a horizontal lift of @xmath7 to the orthonormal frame bundle with antidevelopment @xmath225 . it follows from ( * ? ? ? * proposition 3 ) and novikov s criterion that @xmath0 is a martingale up to time @xmath87 for each @xmath10 . for each @xmath10 we can therefore define a new probability measure @xmath226 on @xmath227 by @xmath228 . it follows from girsanov s theorem that the process @xmath7 when restricted to @xmath229 and considered on the filtered probability space @xmath230 is a fermi bridge between @xmath8 and @xmath2 in time @xmath15 . we therefore obtain a new process , denoted by @xmath23 , defined on @xmath14 and equivalent to the brownian motion @xmath7 on @xmath229 for each @xmath107 . by considering an exhaustion of @xmath0 by regular domains , theorem [ th : secradmomthmbr ] and the monotone convergence theorem imply that it satisfies the bridge property @xmath231 almost surely . in particular , if @xmath2 is a point @xmath232 then one then can extend @xmath23 to a continuous process on @xmath17 $ ] by setting @xmath233 . in this section we prove the main result of this article . denote by @xmath234 the largest domain in @xmath235 with star - like fibres and such that @xmath236 is a diffeomorphism onto its image . then the image of @xmath236 is @xmath221 . recalling that @xmath91 denotes the jacobian determinant of the normal exponential , define also @xmath237 and note that from @xcite there is the formula @xmath238 on @xmath239 . recall also the definition of the geometric local time @xmath240 given in subsection [ ss : firstsec ] by formula . [ th : maintheorem ] suppose that @xmath0 is a complete and connected riemannian manifold of dimension @xmath1 , that @xmath2 is a closed embedded submanifold of @xmath0 of dimension @xmath3 and that @xmath38 is a regular domain in @xmath0 . suppose that @xmath36 with @xmath9 and that @xmath23 is a fermi bridge between @xmath8 and @xmath2 in time @xmath15 . denote by @xmath191 the first exit time of this process from @xmath38 . then , with @xmath241 and @xmath242 defined by and , respectively , we have @xmath243\right]\ ] ] where @xmath244 to begin with , we see by the smooth coarea formula that @xmath245d\xi d\operatorname{vol}_n(z)\\ = & \text { } \lim_{t\uparrow t } \int_n \int_{t_zn^{\perp } } ( p^{d}_t(x,\operatorname{exp})\mathbf{1}_{\mathcal{m}_z(n)}\theta_n)(\sqrt{t - t}\xi)(2\pi)^{-\frac{({m}-{n})}{2}}\exp\left[-\frac{\\|\xi\\|^2}{2}\right]d\xi d\operatorname{vol}_n(z)\\ = & \text { } \int_n \int_{t_zn^{\perp } } ( p^{d}_t(x,\operatorname{exp})\mathbf{1}_{\mathcal{m}_z(n)}\theta_n)(0_z)(2\pi)^{-\frac{({m}-{n})}{2}}\exp\left[-\frac{\\|\xi\\|^2}{2}\right]d\xi d\operatorname{vol}_n(z)\\ = & \text { } \int_n p^{d}_t ( x , z)d\operatorname{vol}_n(z ) \end{split}\ ] ] where @xmath246 denotes the origin of the vector space @xmath247 and where the third equality follows from the compactness of the closure of @xmath38 and the dominated convergence theorem . then , denoting by @xmath248 the dirichlet heat semigroup , it follows from girsanov s theorem that @xmath249 \nonumber \end{split}\ ] ] with @xmath250\ ] ] where @xmath251 is a horizontal lift of @xmath23 to the orthonormal frame bundle and where @xmath225 is the associated @xmath46-valued brownian motion given by the antidevelopment of @xmath251 . now , using it s formula and formula we deduce that @xmath252 and so we can eliminate the stochastic integral in by rearrangement and substitution . finally , using the fact that @xmath253 and also that @xmath254 on @xmath221 together with formula we can further simplify the resulting expression so as to obtain the desired formula . [ th : ihkform ] suppose that @xmath255 is an exhaustion of @xmath0 by regular domains . then , with @xmath256 defined by , we have @xmath257\right]\ ] ] where @xmath25 and @xmath26 are defined by . recalling that @xmath4 is given as the limit of the increasing sequence of dirichlet heat kernels @xmath258 , it follows from the monotone convergence theorem that @xmath259 and so the result follows from theorem [ th : maintheorem ] . the absolutely continuous random measure @xmath25 can be understood precisely in terms of jacobi fields , using heintze and karcher s comparison theorem for @xmath260 which was carefully explained in ( * ? ? ? * section 3 ) . the singular random measure @xmath26 vanishes if the cut locus is polar for @xmath23 . in particular , we have the following corollary of theorem [ th : ihkform ] , which shows that we recover identity in the euclidean setting . [ cor : yarasas ] suppose that @xmath0 is stochastically complete , that the cut locus of @xmath2 has hausdorff dimension at most @xmath186 and that one of the following conditions is satisfied : ( a1 ) : : @xmath3 , the sectional curvature of planes containing the radial direction vanishes and @xmath2 is totally geodesic ; ( a2 ) : : @xmath261 , the ricci curvature in the radial direction vanishes and @xmath2 is minimal . then we have @xmath262 for all @xmath81 and @xmath9 . the corollary follows from theorem [ th : ihkform ] , the heintze - karcher theorem and the monotone convergence theorem . if @xmath2 is a point then theorem [ th : ihkform ] provides a formula for the heat kernel itself . in this case , by the heintze - karcher theorem , the random measure @xmath25 has the property of being non - negative if the ricci curvature of @xmath0 is non - negative and non - positive if the ricci curvature of @xmath0 is non - positive . the heintze - karcher also yields a comparison theorem , stated below , in which we view @xmath263 as a totally geodesic embedded submanifold of @xmath264 and set @xmath265 and @xmath266 . it is a partial generalization of the heat kernel comparison theorem of cheeger and yau @xcite . [ th : ihkcomparison ] suppose that @xmath0 is stochastically complete and that one of the following conditions is satisfied : ( b1 ) : : @xmath3 , the sectional curvature of planes containing the radial direction is bounded below by @xmath267 and @xmath2 is totally geodesic ; ( b2 ) : : @xmath261 , the ricci curvature in the radial direction is bounded below by @xmath268 and @xmath2 is minimal . then we have @xmath269 for any @xmath270 satisfying @xmath271 . first suppose @xmath272 . then @xmath273 , by ( * ? ? ? * theorem 1 ) , so the assertion follows from theorem [ th : ihkform ] since @xmath274 is non - decreasing . so assume @xmath275 and define a function @xmath276 by @xmath277 then , by the heintze - karcher theorem , it follows that @xmath278 since @xmath274 is non - decreasing with @xmath276 non - negative , it follows from this , theorem [ th : ihkform ] and dominated convergence that @xmath279\right].\ ] ] now let @xmath280 be a fermi bridge between @xmath68 and @xmath263 in time @xmath15 such that its radial part @xmath281 satisfies a stochastic differential equation driven by the one - dimensional brownian motion @xmath196 which drives the equation for @xmath282 . then the heintze - karcher theorem , applied to the coefficients of the stochastic differential equation for @xmath283 , combined with the comparison theorem for solutions to stochastic differential equations given in @xcite , implies that @xmath284 , almost surely . since @xmath276 is non - decreasing this yields @xmath285\right].\ ] ] but the right - hand side of is equal to @xmath286 , by theorem [ th : ihkform ] and the fact that @xmath287 so the result follows . if @xmath7 is a brownian motion starting at @xmath81 then @xmath288 is a continuous semimartingale . for the case in which @xmath2 is a hypersurface , the local time of this process at zero is denoted by @xmath289 and referred to as the local time of @xmath7 on @xmath2 . in certain circumstances , such as when @xmath0 is compact , it was explained in @xcite that there is the formula @xmath290 = \int_0^t p^m_s(x , n)\,ds.\ ] ] in the hypersurface case , the integrated heat kernel is therefore given by the time derivative of a mean local time . viewing the @xmath75-dimensional hyperbolic space @xmath76 as an embedded totally geodesic submanifold of @xmath77 , if @xmath7 is a brownian motion in @xmath77 starting at @xmath8 then @xmath291 is a markov process and so , by results in @xcite and a change of variables , formula holds in this setting . therefore , by , it follows that @xmath292 & = \operatorname{sech}(r_{\mathbb{h}^2}(x ) ) ) \int_0^\infty ( 2\pi t)^{-\frac{1}{2 } } \exp\left[-\frac{r_{\mathbb{h}^2}^2(x)}{2t}-\frac{t}{2}\right ] dt\\ & = \operatorname{sech}(r_{\mathbb{h}^2}(x))\exp\left[-r_{\mathbb{h}^2}(x)\right ] . \end{split}\ ] ] in contrast , if @xmath62 is viewed as a linear subspace of @xmath70 with @xmath7 a brownian motion in @xmath70 starting at @xmath8 then @xmath293 = \infty.\ ] ] theorem [ th : ihkcomparison ] yields the following comparison for the mean local time , where we retain the notational convention outlined before theorem [ th : ihkcomparison ] . suppose that @xmath2 is a minimal hypersurface and that the ricci curvature in the radial direction is bounded below by @xmath268 . suppose also that @xmath7 is a non - explosive brownian motion on @xmath0 starting at @xmath8 , denote by @xmath294 the local time on @xmath7 on @xmath2 and assume that formula holds . denote by @xmath295 a brownian motion on @xmath264 starting at @xmath68 with @xmath296 . then @xmath297 \geq \mathbb{e } [ l^{\mathbb{h}^{{n}-1}_\kappa}_t(y(y))]\ ] ] for all @xmath298 . this follows from directly from theorem [ th : ihkcomparison ] and formula . in particular , if @xmath299 with @xmath300 and @xmath301 then the right - hand side of is equal to @xmath302 . the heat kernel lower bounds of cheeger and yau @xcite were proved using a bound on the ricci curvature in the radial direction and a laplacian comparison theorem . these are similar to the objects we use , although our method is quite different . our method is closer in spirit to that of wang @xcite , who also used stochastic techniques with unbounded curvature , but only in the one point case . note that the constants @xmath303 and @xmath304 appearing in the following theorem typically depend upon @xmath2 . [ th : quadcurvlb ] suppose that @xmath0 is stochastically complete and that there exist constants @xmath305 such that one of the following conditions is satisfied off the union of @xmath2 and its cut locus : ( c1 ) : : the sectional curvatures of planes containing the radial direction are bounded below by @xmath306 and there exists a constant @xmath307 such that the principal curvatures of @xmath2 are bounded in modulus by @xmath304 ; ( c2 ) : : @xmath31 and the ricci curvature in the radial direction is bounded below by @xmath308 ; ( c3 ) : : @xmath33 and the ricci curvature in the radial direction is bounded below by @xmath308 and there exists a constant @xmath34 such that the mean curvature of @xmath2 is bounded in modulus by @xmath304 . then for each @xmath9 there exists a constant @xmath309 , depending only on @xmath310 and @xmath158 , such that @xmath311\nonumber\ ] ] for all @xmath81 and @xmath116 $ ] . by ( * theorem 1 ) , the curvature conditions imply @xmath312 with @xmath313 and @xmath314 . using the process @xmath23 constructed in subsection [ ss : lapassump ] it follows from this , theorem [ th : ihkform ] and the fact that @xmath240 is non - decreasing that @xmath315\right]\ ] ] where @xmath316 . for @xmath10 we see that @xmath317&= \mathbf{1}_{\lbrace t < \hat{\tau}_{d_i}\rbrace } \sum_{{p}=0}^{\infty } \frac{{\left(-\int_0^t \frac{f(\hat{x}_s(x))}{t - s}ds\right)^{p}}}{{p}!}\\ & = \mathbf{1}_{\lbrace t < \hat{\tau}_{d_i}\rbrace } -1 + \sum_{{p}=0}^{\infty } \frac{\left(-\mathbf{1}_{\lbrace t < \hat{\tau}_{d_i}\rbrace}\int_0^t \frac{f(\hat{x}_s(x))}{t - s}ds\right)^{p}}{{p}!}\\ & = \mathbf{1}_{\lbrace t < \hat{\tau}_{d_i}\rbrace } -1 + \exp\left[-\mathbf{1}_{\lbrace t < \hat{\tau}_{d_i}\rbrace}\int_0^t \frac{f(\hat{x}_s(x))}{t - s}ds\right ] \end{split}\ ] ] from which it follows , by jensen s inequality , that @xmath318\right]\\ & \quad\quad\quad\quad\quad \geq \mathbb{q}_t \lbrace t < \hat{\tau}_{d_i}\rbrace-1 + \exp\left[-\mathbb{e}\left[\mathbf{1}_{\lbrace t < \hat{\tau}_{d_i}\rbrace}\int_0^t\frac{f(\hat{x}_s(x))}{t - s}ds \right]\right ] . \end{split}\ ] ] inequality implies by formula that inequality holds on @xmath221 with @xmath319 and @xmath320 and so , by theorem [ th : secradmomthmbr ] , we see that @xmath321 & \leq \int_0^t \frac{\mathbb{e}[\mathbf{1}_{\lbrace s<\hat{\tau}_{d_i}\rbrace}r_n^2(\hat{x}_s(x))]}{t - s}ds\\ & \leq \left(r_n^2(x)+\nu t\right)e^{\lambda t } \end{split}\ ] ] for all @xmath10 . similarly , by corollary [ cor : frmebr ] , we have @xmath322 & \leq \int_0^t \frac{\mathbb{e}[\mathbf{1}_{\lbrace s<\hat{\tau}_{d_i}\rbrace}r_n(\hat{x}_s(x))]}{t - s}ds \\ & \leq \left(\frac{r_n^2 ( x ) + \nu t}{t}\right)^{\frac{1}{2}}e^{\frac{\lambda t}{2 } } \int_0^t \left(t - s\right)^{-\frac{1}{2}}ds \\ & = 2 \left(r_n^2 ( x ) + \nu t\right)^{\frac{1}{2 } } e^{\frac{\lambda t}{2 } } \left(1- \left(\frac{t - t}{t}\right)^{\frac{1}{2}}\right)\\ & \leq 2 \left(r_n^2 ( x ) + \nu t\right)^{\frac{1}{2 } } e^{\frac{\lambda t}{2 } } \end{split}\ ] ] for all @xmath10 . consequently @xmath323\leq \alpha \left(r_n^2(x)+\nu t\right)^{\frac{1}{2}}e^{\frac{\lambda t}{2 } } + \frac{\beta}{2}\left(r_n^2(x)+\nu t\right)e^{\lambda t}\ ] ] for all @xmath10 . furthermore @xmath324 by the bridge property and the monotone convergence theorem , so we have a lower bound @xmath325\\ & = q_t(x , n)\exp\left[-({n}\lambda + ( { m}-1)c_1 ) \left(r_n^2(x)+\nu t\right)^{\frac{1}{2}}e^{\frac{\lambda t}{2}}\right . \\ & \quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad \left.-\frac{({m}-1)c_2}{2}\left(r_n^2(x)+\nu t\right)e^{\lambda t}\right ] \end{split}\ ] ] for all @xmath9 and @xmath81 , from which the result follows . since the heat kernel is a positive fundamental solution to the heat equation , fixing one of the spatial variables for small times results in densities whose mass is localized around that fixed point . riemannian manifolds are locally euclidean , so for these small times the resulting densities should be comparable to the gauss - weierstrass kernel . the precise sense in which this is true is given by varadhan s asymptotic relation , originally proved in @xcite , which states for the minimal heat kernel @xmath4 of a complete riemannian manifold @xmath0 that @xmath326 uniformly on compact subsets of @xmath327 . similarly , the embedding of @xmath2 in @xmath0 is locally diffeomorphic to an affine embedding of @xmath73 in @xmath46 so for small times the integrated heat kernel @xmath328 should be comparable to the kernel @xmath329 defined by equation . [ th : genvar ] suppose that @xmath0 is a complete and connected riemannian manifold of dimension @xmath1 and that @xmath2 is a compactly embedded submanifold of @xmath0 of dimension @xmath3 . then @xmath330 uniformly on compact subsets of @xmath0 . it is a simple matter to show that the left - hand side of is less than or equal to the right - hand side , using varadhan s relation and the fact that @xmath331 for all @xmath164 . to prove the opposite inequality assume first that @xmath0 is compact . then the result follows immediately from theorem [ th : quadcurvlb ] . so assume that @xmath0 is non - compact , let @xmath332 be any compact subset of @xmath0 and for @xmath333 and @xmath164 denote by @xmath334 the set of all length - minimizing geodesic segments between @xmath8 and @xmath68 , viewed as a subset of @xmath0 . then @xmath334 contains ( the image of ) at least one such geodesic and by the triangle inequality the set @xmath335 is a bounded subset of @xmath0 . now let @xmath38 be any regular domain in @xmath0 containing @xmath336 . modify @xmath0 outside of @xmath38 so as to obtain a compact riemannian manifold @xmath337 ( by doubling , for example ) and suppose that @xmath38 is sufficiently large so that @xmath338 uniformly for @xmath339 and @xmath164 . this is the principle of _ not feeling the boundary _ considered by hsu in @xcite . such @xmath38 can always be found , as explained by norris in @xcite , since we are assuming that @xmath0 is non - compact . then for all @xmath93 there exists @xmath340 such that for @xmath341 we have @xmath342 for all @xmath333 . it follows from this and the result in the compact case that @xmath343 where @xmath344 denotes the distance function on @xmath337 . but since @xmath336 is contained in @xmath38 it follows that @xmath339 and @xmath164 implies @xmath345 . therefore @xmath346 and the result follows . suppose that @xmath38 is a regular domain . then @xmath347 is a compactly embedded hypersurface and , according to ( * ? ? ? * theorem 5.2.6 ) , one has @xmath348 for all @xmath349 , where @xmath350 denotes the first exit time from @xmath38 of a brownian motion @xmath7 starting at @xmath8 . on the other hand , according to ( * ? ? ? * theorem 1.2 ) , one has @xmath351 for all @xmath352 , where @xmath353 is the measure of heat passing through @xmath38 . theorem [ th : genvar ] and formula imply that , in either case , we have @xmath354 = -\frac{d^2(x,\partial d)}{2}.\ ] ] if @xmath355 is equipped with the sasaki metric @xcite then theorem [ th : genvar ] implies @xmath356 uniformly for @xmath357 in compact subsets of @xmath355 . one could also prove theorem [ th : genvar ] using ndumu s asymptotic expansion @xcite . while this expansion is only valid away from the cut locus , it could be used in place of our lower bounds by connecting points in @xmath332 to @xmath2 with smooth curves , covering them with small balls and invoking the markov property . alternatively , it was proved by hino and ramrez in @xcite , in the context of dirichlet spaces , that @xmath358 for all measurable sets @xmath359 and @xmath225 of positive measure , where @xmath360 denotes the markov process associated with the underlying local regular dirichlet form and where @xmath361 is an associated intrinsic distance . using sturm s upper bounds @xcite and pointwise lower bounds , one can deduce from this the pointwise relation of varadhan , as shown by ramrez ( * ? ? ? * theorem 4.1 ) . a modification of this approach , replacing balls with tubular neighbourhoods and the pointwise lower bounds with our integrated heat kernel lower bounds , might also be used to deduce a relation similar to . grigoryan hu and lau showed in @xcite how gaussian upper bounds on the heat kernel can be obtained from lower bounds . while lower bounds are frequently obtained from upper bounds , as in the method of aronson @xcite , this was the first result to go in the other direction . having proved a lower bound on the integrated heat kernel , and since deducing an upper bound directly from theorem [ th : maintheorem ] seems hard , it is natural for us to use this approach to obtain an upper bound . to do so , we require the following definition . [ eq : lwrreg ] we say that @xmath121 is lower regular if there exist constants @xmath362 such that @xmath363 for all @xmath36 and @xmath364 . for example , if the injectivity radius of @xmath0 is positive and the ricci curvature is bounded above by a constant then @xmath121 is lower regular . for geometric conditions which imply a positive injectivity radius , see @xcite . now assume that @xmath121 is lower regular and that the ricci curvature of @xmath0 is bounded below , by a constant . then the lower bound of theorem [ th : quadcurvlb ] implies a _ near - diagonal lower estimate _ of the form @xmath365 for all @xmath366 and @xmath367 satisfying @xmath368 . this implies , by ( * ? ? ? * corollary 3.5 ) , that there exist constants @xmath369 ( which might depend upon @xmath370 ) such that @xmath371\ ] ] for all @xmath372 and @xmath373 . we therefore have the following theorem , by observing that @xmath164 implies @xmath374 . [ th : grigub ] suppose that the ricci curvature of @xmath0 is bounded below by a constant , that @xmath121 is lower regular and that @xmath2 is compact . then there exist constants @xmath369 ( which might depend upon @xmath370 ) such that @xmath375 \nonumber\ ] ] for all @xmath81 and @xmath376 . we are now in a position to prove gradient and hessian estimates . they are global and hence require assumptions on curvature . for comparison , recall the derivative estimates of cheng , li and yau @xcite . they proved , in particular , that if the injectivity radius of @xmath0 is positive and there exist constants @xmath377 which bound the @xmath378^th^ covariant derivatives of the curvature tensor then for all @xmath9 there exist positive constants @xmath379 and @xmath380 such that @xmath381\ ] ] for all @xmath373 and @xmath382 $ ] . our estimates will be on the _ logarithmic _ derivatives of the _ integrated _ heat kernel . they will be derived using a slight modification of the method used by stroock in @xcite . we will show precisely how the gradient estimate is derived , whereas for the hessian estimate we will refer to @xcite for the details . in particular , to prove the gradient estimate we will need theorem [ th : quadcurvlb ] and a formula for the derivative of the heat semigroup , which we will now state . assume that the ricci curvature of @xmath0 is bounded below , denote by @xmath7 a brownian motion on @xmath0 starting at @xmath8 and denote by @xmath224 a horizontal lift of @xmath7 with antidevelopment @xmath225 . denote also by @xmath383 the solution the ordinary differential equation @xmath384 where @xmath385 . then for any bounded measurable function @xmath386 there is the formula @xmath387\ ] ] for all @xmath42 . this differentiation formula was proved by elworthy and li in @xcite and by li in @xcite , using basic stochastic calculus , while another , related , approach was given by thalmaier in @xcite . it was proved originally by bismut in @xcite for the case in which @xmath0 is compact . an analogous formula was used by stroock , whose estimate is based upon the following lemma , proved as ( * ? ? ? * lemma 6.45 ) using jensen s inequality . [ lem : stroocklem ] suppose @xmath388 is a probability space and @xmath389 a non - negative measurable function on @xmath390 with @xmath391=1 $ ] . if @xmath389 is a measurable function on @xmath390 such that @xmath392 is integrable , then @xmath393 \leq \mathbb{e}\left[\phi \log \phi\right]+\log \mathbb{e}[e^\psi]$ ] . [ th : derivativeestiamte ] suppose that @xmath0 is a complete and connected riemannian manifold of dimension @xmath1 and that @xmath2 is a compactly embedded submanifold of @xmath0 of dimension @xmath394 . suppose that the injectivity radius of @xmath0 is positive and that the curvature tensor is bounded . then for all @xmath9 there exists a constant @xmath395 such that @xmath396 for all @xmath81 and @xmath116 $ ] . for a bounded measurable positive function @xmath40 and @xmath397 set @xmath398 and see , by lemma [ lem : stroocklem ] and formula , that @xmath399\right ] \nonumber\ ] ] where @xmath400.\ ] ] furthermore , denoting by @xmath401 the minimum of the ricci curvature , equation implies @xmath402\right ] \leq \frac{\gamma^2}{2}\int_0^t e^{-rs}ds\ ] ] and so , after minimizing over @xmath403 , we deduce @xmath404 now we choose @xmath405 . then @xmath406 , by tonelli s theorem , and for all @xmath407 it follows that @xmath408 the curvature assumptions imply that there exist constants @xmath305 such that the sectional curvatures of planes containing the radial direction are bounded below by @xmath409 and so , by theorem [ th : quadcurvlb ] , there exists a constant @xmath410 , depending only on @xmath411 and @xmath158 , such that @xmath412\ ] ] for all @xmath81 and @xmath116 $ ] . the assumptions also imply , by theorem [ th : grigub ] and the chapman - kolmogorov equation , that there exists a constant @xmath413 such that @xmath414 for all @xmath116 $ ] and @xmath407 . substituting the estimates and in to proves the theorem . it follows that the gradient estimate holds automatically if @xmath0 is compact . furthermore , we can now prove the semimartingale property promised in subsection [ ss : smprop ] . as pointed out by thalmaier , the semimartingale property should hold without any assumptions on curvature . gneysu proved this for the one point case using local estimates on the heat kernel . we do not yet have such estimates for the integrated heat kernel and so , for the time being , we make do with the assumptions of theorem [ th : derivativeestiamte ] . it suffices to control the singularity in the drift close to the terminal time . since the distance function @xmath188 is defined as an infimum over @xmath2 , it follows from and lemma [ lem : concrit ] that there exists @xmath145 such that @xmath416 \leq c_\epsilon(t - t)$ ] for all @xmath417 $ ] . therefore , by theorem [ th : derivativeestiamte ] , there exists @xmath395 such that @xmath418\\ \leq & \text { } \int_{t-\epsilon}^{t } \mathbb{e}^{x;n , t}\left[\\| \nabla \log p^m_{t - t}(x_t , n)\\|^2\right]^{\frac{1}{2 } } dt\\ \leq & \text { } \sqrt{c}\int_{t-\epsilon}^{t } \left(\frac{1}{t - t}+\frac{{n}}{t - t}\log\frac{1}{t - t}+\frac{\mathbb{e}^{x;n , t}\left[r^2_n(x_t)\right]}{(t - t)^2}\right)^{\frac{1}{2 } } dt\\ \leq & \text { } \sqrt{c}\int_{t-\epsilon}^{t } \left(\frac{1}{t - t}+\frac{{n}}{t - t}\log\frac{1}{t - t}+\frac{c_\epsilon}{t - t}\right)^{\frac{1}{2 } } dt\\ < & \text { } \infty \end{split}\ ] ] and the result follows . note that we could have derived our gradient estimate using assumptions slightly weaker than those of theorem [ th : derivativeestiamte ] , using ricci curvature , the sectional curvature of planes containing the radial direction and lower regularity of the volume measure . to derive our hessian estimate we need only add to these assumptions some suitable control on the curvature two - form and the derivative of the ricci tensor . for simplicity , however , we state our hessian estimate , as we did the gradient estimate , in terms of injectivity radius and the full curvature tensor . this way , our estimates can be easily compared to that of cheng , li and yau , stated above for the case in which the injectivity radius of @xmath0 is positive . [ cor : hesscor ] in addition to the assumptions of theorem [ th : derivativeestiamte ] suppose also that the first covariant derivative of the curvature tensor is bounded . then for all @xmath9 there exists a constant @xmath395 such that @xmath419 for all @xmath81 and @xmath116 $ ] . it was proved by stroock in @xcite that for any continuous positive function @xmath40 there exists @xmath395 such that @xmath420 for all @xmath81 and @xmath421 $ ] where @xmath422 is defined by . choosing @xmath423 , using the lower bound and the on - diagonal upper bound we obtain the corollary , by theorem [ th : derivativeestiamte ] and the fact that @xmath424 for all @xmath425 . note that the constant @xmath426 appearing in this estimate depends only on the injectivity radius of @xmath0 , the length of the time interval @xmath15 , the dimensions @xmath1 and @xmath158 and the bounds on the curvature tensor and its derivative . the results in this article , in particular the estimate on the hessian , will lead to a study of the space of continuous paths which end on a submanifold , potentially shedding light on the relationship between the geometry of the path space , the intrinsic geometry of the ambient manifold and the extrinsic geometry of the submanifold , extending recent work of naber @xcite to the submanifold setting . a related application , which also concerns the extrinsic geometry of the submanifold , is the study of mean curvature flow ( mcf ) . mcf typically concerns the codimension one @xmath427 case , where the normal component of the velocity of a point on a hypersurface @xmath2 is given by the mean curvature at that point . an important object here is the @xmath12-functional @xmath428 introduced by huisken @xcite ( and for which theorem [ th : ihkform ] now provides a probabilistic formula ) . the monotonicity of this functional under mcf is the reason for its importance . the _ entropy _ @xmath429 of the hypersurface ( see ( * ? ? ? * section 3 ) ) is then defined by @xmath430 . the entropy captures something of the extrinsic complexity of @xmath2 and determines which singularities can form as a result of the flow ( and by theorem [ th : quadcurvlb ] , the entropy is always greater than or equal to @xmath180 ) . according to subsection [ ss : hltcomp ] , one can interpret the @xmath12-functional as the ratio @xmath431}{\frac{d}{dt}\mathbb{e}[l^{\mathbb{r}^{{m}-1}}_t(x(0))]}.\ ] ] indeed , the rate at which the expected local time increases for a hypersurface with high mean curvature ought to be greater than that for , say , a minimal hypersurface . so mcf has an interpretation in terms of hypersurface local time which could lead to further applications of our results . less abstractly , note that any non - degenerate it diffusion in @xmath46 , constrained to arrive in a submanifold at a fixed positive time , such as on a level set of a function or on the boundary of a domain , is a process of the type considered in this article ( any non - degenerate diffusion on a smooth manifold induces a riemannian metric , via the principle symbol of its generator , with respect to which it is a brownian motion with drift ) . multi - dimensional it diffusions constrained by the value of an observable at a fixed future time could have applications in finance . alternatively , consider the case of a particle diffusing inside a biological cell and which is known to exit the cell , via the membrane , at a particular time . then the possible trajectories of the particle are distributed as those of the corresponding conditioned diffusion ( and the metric could even be used to capture some of the internal structure of the cell ) . finally , a connection to statistics is given by the problem of manifold denoising ( see @xcite ) , in which data is drawn from sampling a probability distribution supported around a submanifold of euclidean space . the idea is then to reconstruct the submanifold ( or derive certain topological properties of it , as in @xcite ) . the integrated heat kernel @xmath432 ( normalized by the hypersurface volume @xmath433 ) yields a probability distribution corresponding to the case in which observations of the submanifold are made subject to a gaussian error . jeff cheeger , mikhail gromov , and michael taylor , _ finite propagation speed , kernel estimates for functions of the laplace operator , and the geometry of complete riemannian manifolds _ , j. differential geom . * 17 * ( 1982 ) , no . 1 , 1553 . mr658471 ( 84b:58109 ) michael cranston , wilfrid s. kendall , and peter march , _ the radial part of brownian motion . ii . its life and times on the cut locus _ , probab . theory related fields * 96 * ( 1993 ) , no . 3 , 353368 . mr1231929 ( 94g:60154 ) bruce k. driver , _ a cameron - martin type quasi - invariance theorem for pinned brownian motion on a compact riemannian manifold _ , trans . * 342 * ( 1994 ) , no . 1 , 375395 . mr1154540 ( 94e:60072 ) k. d. elworthy and a. truman , _ the diffusion equation and classical mechanics : an elementary formula _ , stochastic processes in quantum theory and statistical physics ( marseille , 1981 ) , lecture notes in phys . , 173 , springer , berlin , 1982 , pp . mr729719 ( 85g:58090 ) pat fitzsimmons , jim pitman , and marc yor , _ markovian bridges : construction , palm interpretation , and splicing _ , seminar on stochastic processes , 1992 ( seattle , wa , 1992 ) , progr . 33 , birkhuser boston , boston , ma , 1993 , pp . mr1278079 ( 95i:60070 ) ernst heintze and hermann karcher , _ a general comparison theorem with applications to volume estimates for submanifolds _ , ann . sci . cole norm . ( 4 ) * 11 * ( 1978 ) , no . 4 , mr533065 ( 80i:53026 ) martin ngu ndumu , _ the heat kernel formula in a geodesic chart and some applications to the eigenvalue problem of the @xmath50-sphere _ , probab . theory related fields * 88 * ( 1991 ) , no . 3 , 343361 . mr1100896 ( 93a:58177 ) shige peng and xuehong zhu , _ necessary and sufficient condition for comparison theorem of 1-dimensional stochastic differential equations _ , stochastic process . * 116 * ( 2006 ) , no . 3 , 370380 . mr2199554 o. g. smolyanov , h. v. weizscker , and o. wittich , _ brownian motion on a manifold as limit of stepwise conditioned standard brownian motions _ , stochastic processes , physics and geometry : new interplays , ii ( leipzig , 1999 ) , cms conf . , vol . 29 , amer . math . soc . , providence , ri , 2000 , pp . mr1803450 ( 2001k:58078 ) daniel w. stroock , _ an introduction to the analysis of paths on a riemannian manifold _ , mathematical surveys and monographs , vol . 74 , american mathematical society , providence , ri , 2000 . mr1715265 ( 2001m:60187 ) karl - theodor sturm , _ analysis on local dirichlet spaces . ii . upper gaussian estimates for the fundamental solutions of parabolic equations _ , osaka j. math . * 32 * ( 1995 ) , no . 2 , 275312 . mr1355744 ( 97b:35003 ) keith d. watling , _ formul for the heat kernel of an elliptic operator exhibiting small - time asymptotics _ , stochastic mechanics and stochastic processes ( swansea , 1986 ) , lecture notes in math . 1325 , springer , berlin , 1988 , pp . mr960168 ( 90j:58146 ) keith d. watling , _ formul for solutions to ( possibly degenerate ) diffusion equations exhibiting semi - classical asymptotics _ , stochastics and quantum mechanics ( swansea , 1990 ) , world sci . publ . , river edge , nj , 1992 , pp . mr1182603 ( 93f:35107 )	we introduce and study brownian bridges to submanifolds . our method involves proving a general formula for the integral over a submanifold of the minimal heat kernel on a complete riemannian manifold . we use the formula to derive lower bounds , an asymptotic relation and derivative estimates . we also see a connection to hypersurface local time . this work is motivated by the desire to extend the analysis of path and loop spaces to measures on paths which terminate on a submanifold . + * keywords : * brownian bridge ; local time ; heat kernel ; submanifold * ams msc 2010 : * 58j65 ; 53b21 ; 60j55
the inverse problem is described in figure [ pbinv ] . the radar cross section quantifies the scattering power of an object , at a given incidence and wave frequency . it is defined as the ratio between the radar transmitted power and the incident power density ( in plane wave ) @xcite . rcs measurement process is schematically presented on the right part of figure [ pbinv ] . the object or mock - up is illuminated by a quasi - planar monochromatic wave , inside an anechoic chamber where interferences are limited.the acquisitions are realized at @xmath0 successive discrete frequencies ( @xmath1 ) , for different incidence angles ( by piloting motorized rotating support ) . from this raw data , a signal processing is performed , mainly consisting of calibration and filtering . at the end , the measurement provides an evaluation of the calibrated complex ( amplitude and phase ) scattering coefficient , for each frequency and incidence . a metallic axi - symmetric object is recovered with @xmath2 areas ( see figure [ freq - fix ] ) , each area corresponding to a material with its associated isotropic radioelectric properties , i.e. the complex parameters of permittivity @xmath3 and permeability @xmath4 . the em inverse problem can be expressed as : is it possible to extract some local information on the material properties @xmath5 of each area from the global scattering measurement ? at a given frequency @xmath6 , the system state @xmath7 can be defined by ( omitting @xmath6 to lighten the notations ) : @xmath8^t $ ] , with @xmath9 , @xmath10 , @xmath11 and @xmath12 , respectively the real and imaginary permittivity and permeability components of the @xmath2 areas . notice that the state dimension can be high ( @xmath13 with @xmath14 ) . on the other hand , considering measurements at the given frequency @xmath6 with various incidence angles in both polarizations ( tm and te ) , the observation vector is made of the real ( @xmath15 ) and imaginary ( @xmath16 ) parts of the complex scattering coefficients @xmath17 and @xmath18 measured at @xmath19 angles @xmath20 : @xmath21^t $ ] . considering all frequencies , vectors @xmath22 and @xmath23 respectively define complete system state and observation . the 2d - axisymetric maxwell solver software ( @xmath24 ) can predict the observation from the system state . assuming a multidimensional gaussian measurement uncertainty model , it leads to the following likelihood model @xmath25 ( at a given frequency @xmath6 ) , where @xmath26 is the covariance matrix ( assumed known ) . to avoid numerous and heavy @xmath24 computations , we have developed the following global approach . first , the high dimension state space and the associated system response are explored randomly around expected properties ( prior knowledge ) , computations being massively distributed on hpc machines . let @xmath27 the training data composed of @xmath28 couples . it is then processed by n - d statistical techniques ; sensibility analysis and model reduction techniques can possibly reduce the state space dimension . applying multidimensional regression , it turns out that the model is approximatively linear . according to the studied cases , the linearity errors , evaluated by residue analysis and bootstrap techniques , are significant but much lesser than the rcs measurement uncertainties . finally , the following linear gaussian model can be considered ( at a given frequency @xmath6 ) : @xmath29 where the deterministic part of the linear model is given by the learned matrix @xmath30 @xmath31 and the vector @xmath32 . let us fix a frequency @xmath6 . we model our a priori knowledge on @xmath7 with a gaussian distribution @xmath33 . the object is divided in @xmath34 blocks of areas , each of them composed of a rather homogeneous material . the location of these blocks is known exactly . prior mean value @xmath35 is defined with reference values of @xmath36 , @xmath37 , @xmath38 and @xmath39 for each of theses blocks . then , for any component @xmath40 , we define a variance @xmath41 as a mix between absolute and relative uncertainty . to take into account the spatial local homogeneity , covariance matrix @xmath42 is defined block by block independently ( and separately for each of the @xmath36 , @xmath37 , @xmath38 , @xmath39 ) by correlation relations between @xmath7 s block - sharing components : @xmath43 where @xmath44 $ ] is a spatial correlation parameter ( typically @xmath45 in our applications ) . it means that the correlation between @xmath46 areas dicreases geometrically at speed @xmath47 with the distance between them . radioelectric properties @xmath36 , @xmath37 , @xmath38 , @xmath39 are known to vary in function of the wave frequency @xcite ; their non - stationary dynamic can be quite different from frequency @xmath48 to @xmath49 . however , in order to take account of expected frequency profiles regularity ( for each material and @xmath36 , @xmath37 , @xmath38 , @xmath39 ) , we model sequence @xmath50 with a linear gaussian correlation structure , given by the generalised autoregressive ( ar ) random process @xcite : @xmath51 where * for each @xmath52 , @xmath53 is the `` square root '' matrix of covariance matrix @xmath42 , defined as being the unique symetric definite positive matrix satisfying @xmath54 ; * @xmath55 are i.i.d . @xmath56 ; * @xmath57 is a positive diagonal matrix commuting with the @xmath53 , further described , and depending on a parameter @xmath58 . one notices that so defined sequence @xmath7 still admits @xmath33 as marginal distributions . more generally , it can be shown that the distribution of concatenated vector @xmath59 is a gaussian distribution with mean @xmath60 and covariance matrix : @xmath61 where @xmath62 is the diagonal by block matrix @xmath63 , which explicits every joint distribution @xmath64 . + finally , let us clarify parameter @xmath58 ( and matrix @xmath57 ) . @xmath57 s role in these equations makes clear that it s a frequential correlation parameter . its dimension is to be chosen among @xmath65 possibilities according to our assumptions . * : @xmath58 is assumed independant from material and @xmath36 , @xmath37 , @xmath38 , @xmath39 . it is one - dimensional , @xmath66 $ ] , and @xmath67 . * : @xmath58 depends on the material ( block ) . it is @xmath34-dimensional , @xmath66^{n_b}$ ] , and @xmath57 giving to each line ( component of @xmath7 ) its associated component of @xmath58 . * : @xmath58 depends on material and @xmath36 , @xmath37 , @xmath38 , @xmath39 . it is @xmath68-dimensional , and @xmath57 is the diagonal matrix composed with @xmath69 @xmath70-type blocks . conditionnaly to frequential correlation parameter @xmath58 , the problem of the determination of @xmath71 given the measurements @xmath72 can be expressed as a classic linear gaussian hidden dynamic markov process observed at `` times '' @xmath6 ( @xmath73 ) : @xmath74 where @xmath75 and @xmath76 are independant gaussian noises with known parameters , and @xmath77 a known matrix depending on @xmath58 ( see ( [ mod - obs ] ) and ( [ mod - ap ] ) ) . parameter @xmath58 , intuitively representing the inter - frequency regularity , is unknown and to be estimated . in respect to the probabilistic point of vue , it is probabilized , and given a prior distribution @xmath78 . the posterior distribution @xmath79 can be decomposed as : @xmath80 . conditionally to @xmath58 , the system is linear gaussian : the conditional distributions @xmath81 can be straightforwardly computed by kalman filtering , including in this off - line context backward kalman smoothing . on the other hand , the term @xmath82 can be evaluated ( up to a normalising constant ) for a given @xmath58 using the likelihood term provided by the kalman filter and prior distribution @xmath78 . consequently , in order to exploit this conditional structure of the system , kalman smoothers are applied and integrated in an interacting particle approach . this idea of mixing analytic integration ( here kalman evaluation of @xmath83 ) with stochastic sampling is a variance reduction approach , known as rao - blackwellisation @xcite . similarly to @xcite , we choose to implement an efficient interacting particle approach , in order to estimate the marginal distribution @xmath84 . sequential monte carlo ( smc ) is a stochastic algorithm to sample from complex high - dimensional probability distributions . the principle ( e.g. , @xcite ) is to approximate a sequence of target probability distributions @xmath85 by a large cloud of random samples termed particles @xmath86 , @xmath87 being called the state space . between `` times '' @xmath88 and @xmath89 , the particles evolve in state space @xmath87 according to @xmath46 steps : * a * selection * step : every particle @xmath90 is given a weight @xmath91 defined by a selection function @xmath92 ( @xmath93 ) . by resampling ( stochastic or deterministic ) , low - weighted particles vanish and are replaced by replicas of high - weighted ones . * a * mutation * step : each selected particle @xmath94 move , independently from the others , according to a markov kernel @xmath95 . @xmath96{cl}% \zeta_{n-1}^{1 } & \\ \vdots & \\ \zeta_{n-1}^{i } & \\ \vdots & \\ \zeta_{n-1}^{n_p } & \end{array } \right ] \underbrace { \overset{g_n } % { -\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\longrightarrow } } _ { \textmd{selection}}\left [ \begin{array } [ c]{cl}% \widehat{\zeta}_{n-1}^{1 } & \overset{m_{n}}% { -\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\longrightarrow } \\ \vdots & \\ \widehat{\zeta}_{n-1}^{i } & -\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\longrightarrow \\ \vdots & \\ \widehat{\zeta}_{n-1}^{n_p } & \underbrace{-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\longrightarrow}_{\textmd{mutation } } \end{array } \right . \left . \begin{array } [ c]{cl}% \zeta_{n}^{1 } & \\ \vdots & \\ \zeta_{n}^{i } & \\ \vdots & \\ \zeta_{n}^{n_p } & \end{array } \right]\ ] ] evolving this way , the cloud of particles , and more precisely the occupation distribution @xmath97 ( sum of dirac distributions ) , approximates for each @xmath89 the theoretical distribution @xmath98 defined recursively by the feynman - kac formulae , associated with the potentials @xmath92 and kernels @xmath95 @xcite . + back to our objective of sampling from @xmath99 , we then define the sequence of distributions @xmath100 : @xmath101 where @xmath102 is a sequence of number increasing from @xmath103 to @xmath104 , so that : @xmath105 is prior distribution @xmath78 , easy to sample , @xmath106 is target distribution @xmath99 and sequence @xmath85 admits a feynman - kac type structure with calculable selection functions @xmath92 and markov kernels chosen so that @xmath107 ( metropolis - hastings for example ) . the distribution @xmath99 is then interpreted as being the last distribution of a feynman - kac sequence , on which smc can be performed , the estimator of @xmath99 being the occupation distribution @xmath108 extracted from the last cloud of particles . the occupation distribution @xmath108 which approximates @xmath109 can be represented dimension by dimension via histograms ( see figure [ hist - rau ] ) . for each frequency @xmath6 , this approximation @xmath110 , associated with the theoretical conditioning relations @xmath111 \\ & cov(\mathbf{x}_k \| \mathbf{y } ) = e\left [ cov(\mathbf{x}_k \| \rho,\mathbf{y } ) \| \mathbf{y } \right ] + cov\left [ e(\mathbf{x}_k \| \rho,\mathbf{y } ) \| \mathbf{y } \right]\end{aligned}\ ] ] can deliver estimators of respectively the mean and the covariance matrix of @xmath7 .roughly speaking , the posterior estimation is performed by randomly picking a @xmath112 from the final cloud of particles and computing associated samples of @xmath7 by a kalman smoother conditionally to @xmath112 . it is illustrated in figure [ freq - fix ] , with a good agreement between the true state and estimated state . moreover , for any fixed area , the method provides estimators of the mean and marginal variance for every frequency , so that the results can be presented as frequential profiles , with marginal uncertainties ( see figure [ ann - fix ] ) . even when the true ( simulated ) em property profiles are chosen markedly divergent from the prior ar - type model , the method turns out to be robust . it results from the adaptive estimation of @xmath58 which provides small values of @xmath58 ( i.e. weak correlation of em properties for close frequencies ) in the case of highly irregular true profiles . an efficient statistical inference approach has been applied to estimate local material radioelectric properties from global em scattering measurements . it combines intensive computations , meta - modeling and advanced sequential monte carlo techniques dedicated to frequency dynamic estimation . 9 e. f. knott , _ radar cross section measurements _ , scitech publishing , 2006 . s. m. pandit , s - m wu , _ time series and system analysis with applications _ , john wiley & sons , 1983 . j. s. liu and r. chen , _ sequential monte carlo for dynamic systems _ , journal of the american statistical association , 93 , 1032 - 1044 , 1998 . p. del moral , a. doucet and a. jasra , _ sequential monte carlo methods for bayesian computation _ , bayesian statistics 8 , oxford university press , 2006 . del moral , _ feynman - kac formulae , genealogical and interacting particle approximations _ , springer new york ( series : probability and applications ) , 2004 .	the following electromagnetism ( em ) inverse problem is addressed . it consists in estimating local radioelectric properties of materials recovering an object from the global em scattering measurement , at various incidences and wave frequencies . this large scale ill - posed inverse problem is explored by an intensive exploitation of an efficient 2d maxwell solver , distributed on high performance computing ( hpc ) machines . applied to a large training data set , a statistical analysis reduces the problem to a simpler probabilistic metamodel , on which bayesian inference can be performed . considering the radioelectric properties as a dynamic stochastic process , evolving in function of the frequency , it is shown how advanced markov chain monte carlo methods , called sequential monte carlo ( smc ) or interacting particles , can provide estimations of the em properties of each material , and their associated uncertainties .
double layer two - dimensional ( 2d ) carrier systems as realized in semiconductor nanostructures are of interest with respect to fundamental phenomena as fractional quantum hall effect and superconductivity because of the interlayer carrier interaction @xcite . depending on the width of the potential barrier that separates the two layers one has to distinguish the regime of quantum - mechanical coupling via tunneling for small barrier widths from that of interlayer coupling mediated by ( direct or effective ) electron - electron interaction for sufficiently large barriers . in the former case , the frictional aspects of superconductivity @xcite and of the fractional quantum hall effect @xcite have been studied . in the latter case , when the large barrier allows independent control of electron conduction in the two layers , the frictional drag effect manifests itself when a current driven along one layer ( layer @xmath2 ) induces , via momentum transfer , a drag voltage in the second layer ( layer @xmath3 ) under conditions that no current flows in this well @xcite . the drag effect has been studied experimentally for two 2d electron systems ( 2des ) , two 2d hole systems ( 2dhs ) , but also in situations with layer @xmath2 @xmath3 being a 2des or 2dhs , respectively @xcite . recently the effect of a magnetic field perpendicular to the layers has been investigated @xcite . theoretical work on the frictional drag effect was devoted to elucidating the microscopic mechanism of the momentum transfer @xcite . the calculated temperature dependence of the frictional drag due to direct coulomb electron - electron interaction shows strong dependence on the inter - layer separation @xmath4 @xcite while in the experiment by gramila et al . @xcite , the temperature dependence of the observed frictional drag for separations at least up to @xmath5 nm exhibits almost no dependence on @xmath4 , and has been interpreted as being due to the exchange of virtual phonons . the frictional drag accomplished by the exchange of virtual phonons has been studied theoretically in . this work , however , because of its conceptual complexity requires involved numerical computations and does not lead to analytical expressions for the dependence of the frictional drag on the system parameters . therefore , we present here an analytical derivation of the drag scattering rate for the particular case of asymmetric two - layer systems ( @xmath6 ) coupled by virtual exchange of acoustic phonons including its temperature dependence . we thus model the experimental situation investigated by t. j. gramila , et al . @xcite and find by using the corresponding system parameters the characteristic @xmath7-dependence of the drag scattering rate in quantitative agreement with these experimental data . our model as depicted in fig . [ fg1 ] consists of two parallel electron sheets separated by a distance @xmath4 with electron densities @xmath8 and @xmath9 and the layer extensions @xmath10 and @xmath11 . according to experimental observations , a current @xmath12 driven along the layer @xmath2 by applying a voltage @xmath13 leads by frictional drag to a voltage @xmath14 in the layer @xmath3 . the drag scattering rate in a double layered system can be defined as @xcite @xmath15 where @xmath16 are the electric fields connected with the voltages @xmath17 , and @xmath18 is the transport relaxation time which determines via the mobility @xmath19 the current in the layer @xmath2 ; @xmath20 is the electron effective mass . following , we obtain by using the linearized boltzmann equation for the drag scattering rate @xmath21 where @xmath22 is the normalization length , @xmath23 the electron in - plane momentum , @xmath24 the electron spin , @xmath25 the transferred momentum , @xmath26 the fermi distribution function determined by temperature @xmath7 and fermi energy @xmath27 with fermi wave vector related to the areal electron concentration @xmath28 , by @xmath29 . @xmath30 is the transition probability of two electrons from the states @xmath31 and @xmath32 with energies @xmath33 and @xmath34 , respectively , into the states @xmath35 and @xmath36 with energies @xmath37 and @xmath38 . the index @xmath39 refers to the type of effective electron - electron interaction which causes this transition . in this work we consider effective electron - electron interaction mediated by the exchange of virtual acoustic phonons related to the piezoelectric ( @xmath40 pa ) and deformation ( @xmath40 da ) potential coupling . using fermi s golden rule , @xmath30 can be written in the form @xmath41 where @xmath42 is the transition matrix element . phonon mediated electron - electron interaction appears in second order perturbation theory with respect to the electron - phonon coupling . the transition matrix element of this process is visualized by the diagram in fig . [ fg2 ] . for intrasubband scattering the vertex part of this diagram corresponds to @xmath43 with the subband electron density @xmath44 and @xmath45 . @xmath46 depends only on the absolute value of @xmath47 @xcite @xmath48 here we describe electron - phonon interaction in the framework of the isotropic model with @xmath49 being the longitudinal sound velocity , @xmath50 the crystal mass density , @xmath51 and @xmath52 are the piezoelectric and deformation potential constants , respectively , averaged over the directions of the vector @xmath53 and the phonon polarizations @xcite . we assume also that all elastic parameters of the sample are the same so that phonons are not reflected from the interfaces separating different materials ( for instance between gaas and algaas ) . the phonon propagator is @xmath54 with @xmath55 , @xmath56 and a summation over phonon momenta @xmath53 corresponds to the internal phonon line ( the dashed line in the diagram ) . in the above expression of the phonon propagator , @xmath57 is the lifetime of the intermediate phonon states with respect to all processes which are destroying these intermediate states . for virtual intermediate states there is no energy conservation at the vertices of the diagram , _ i.e. _ @xmath58 can be different from @xmath59 and we may assume infinite lifetime @xmath57 for the intermediate phonons when calculating the drag rate . then all integrals with respect to the energy parameters in eq . ( [ eq2 ] ) are to be understood in the sense of the principal value . taking the summation over intermediate momenta @xmath53 , we obtain for the squared - modulus of the transition matrix elements for the two considered mechanisms @xmath60 where @xmath61 ^ 2 } , \label{eq10 } \\ & & \|i^{da}(\alpha,\lambda)\|^2={e^{-2\alpha\lambda}\over(\alpha\lambda)^2}\ ; { ( \omega\lambda / s)^4\;\sinh^2(\alpha d/2)\over ( \alpha d/2)^2\left[1+(\alpha d/2\pi)^2\right]^2 } \label{eq11}\end{aligned}\ ] ] and @xmath62 is defined generally as @xmath63 where @xmath64 is the positive infinitesimal , and the branch cut for the square root is assumed to lie along the negative real axis . it is easy to check that with such definition of @xmath62 , the expressions of the form factors ( [ eq10 ] ) and ( [ eq11 ] ) are valid both for @xmath65 and @xmath66 . in calculating the form factors @xmath67 and @xmath68 we have assumed that electrons are localized in symmetric infinitely high quantum wells with the width @xmath69 , thus having @xmath70 and @xmath71 as explicit forms for the electron density functions in layers @xmath2 and @xmath3 , respectively . in eqs . ( [ eq8 ] ) and ( [ eq9 ] ) we introduce also nominal scattering times @xmath72 with the numerical values obtained for a gaas quantum well with the electron concentration @xmath73 @xmath74 as in . substituting eqs . ( [ eq3]),([eq8 ] ) , and ( [ eq9 ] ) into eq . ( [ eq2 ] ) and taking integrals over @xmath75 and over directions of the vector @xmath76 by exploiting @xmath77-functions , we can represent the drag scattering rate in the form @xmath78 where @xmath79 in this equation @xmath80 and the limiting energy @xmath81 is obtained from the energy and momentum conservation laws . now making use of the identity @xmath82 and taking integrals in eq . ( [ eq15 ] ) with respect to the energy parameters , we reduce the expression for @xmath83 to a one - dimensional integral of the form @xmath84 where @xmath85 $ ] gives the polylogarithm function of the order @xmath86 @xcite @xmath87\equiv li_\nu(z)= { z\over \gamma(\nu ) } \int_0^\infty dt\ , { t^{\nu-1}\over e^t - z } \label{eq18}\ ] ] and @xmath88 for the special gaas system with @xmath89 @xmath74 we have @xmath90 k)@xmath91 and @xmath92 @xmath93 . one can see from eq . ( [ eq17 ] ) that in the framework of the adopted approximation when the lifetime of the intermediate phonon states is infinite , _ i.e. _ @xmath94 , the drag scattering rate , eq . ( [ eq14 ] ) , is independent of the inter - layer spacing @xmath4 . this means that our theory refers to the experimental situation when @xmath4 is smaller than the phonon mean free path , @xmath95 , associated with the finite phonon lifetime . for @xmath96 , we have to consider the finite @xmath97 in the phonon propagator , which would lead to an exponential dependence of the scattering probability of @xmath98 . using eqs . ( [ eq14 ] ) and ( [ eq17 ] ) we plot the temperature dependence of the drag scattering rate divided by @xmath0 for identical layers in a symmetric gaas / algaas double quantum well structure as in coupled both by pa- and da - phonon mediated electron - electron interaction ( fig . [ fg3 ] ) . it is seen from this figure that the contribution of the da interaction is smaller ( approximately by one order of magnitude ) than that of the pa interaction . this is because the scattering probability due to long - range piezoelectric interaction includes an additional factor @xmath99 with respect to the short - range deformation potential interaction . this difference is strongly pronounced at low temperatures @xcite where the pauli exclusion principle restricts electron - phonon scattering processes in a fermi gas to those with small angles . our calculations correspond to the experimental situation of ref . where the drag rate was measured for samples with the inter - layer spacing @xmath100 , and @xmath101 nm with the electron concentration @xmath102 @xmath74 in a temperature range from @xmath2 to @xmath103 k. starting from the lower edge of this interval , where the drag rate is small because it is dominated by small angle scattering connected with a small momentum transfer , the drag rate increases with increasing temperature . towards the upper edge of the interval , large angle scattering becomes dominant which reduces the increase of the drag scattering rate in temperature . this gives rise to a peak of the function @xmath104 both for drag caused by the exchange of da- and pa - phonons at approximately the same temperatures as shown in fig . [ fg3 ] . the maximum drag rate of about @xmath2 s@xmath105 k@xmath106 due to combined effect of da- and pa - phonons is obtained at a temperature near @xmath107 k in good agreement with the experimental results for the samples with the inter - layer spacing @xmath108 nm @xcite . our theory is not applicable for much larger interlayer spacing ( as _ e.g. _ for the sample with @xmath109 nm in ref . ) . we believe that in this sample due to other scattering mechanisms , the intermediate virtual phonon states which effectively realize electron - electron interaction , are destroyed and this leads to the exponential suppression of the drag scattering rate observed in the experiment . in fig . [ fg4 ] we present the temperature dependence of the drag scattering rate for a symmetric double quantum well system as in fig . [ fg3 ] but for different values of the matched electron density @xmath28 ( the second line from the top ( @xmath102 @xmath74 ) is identical with solid line in fig . [ fg3 ] . ) . with increasing @xmath28 the peak structure becomes broader while the maximum value decreases and the position @xmath110 shifts to higher temperature . this shift follows approximately the relation @xmath111 ( see the inset of fig . [ fg4 ] ) which supports our findings that the peak position is related to the energy @xmath112 which separates the regions of small ( @xmath113 ) and large ( @xmath114 ) angle scattering . we have also studied the drag scattering rate as in fig . [ fg3 ] with fixed density @xmath115 @xmath74 but changing density @xmath9 at different temperatures . a peak occurs always at matched densities for which the momentum transfer is most efficient . it is seen for @xmath116 k and most clearly pronounced at @xmath117 k while for @xmath118 k it is only a small feature on an otherwise monotonous function of @xmath9 . this strikingly different behavior for @xmath118 k as compared to @xmath117 and @xmath119 k which is similar to the experimental finding in refs . , can be attributed to the changing dominance of small and large angle scattering with temperature and @xmath9 . in conclusion , we have calculated the drag scattering rate between spatially separated electron layers due to the virtual phonon exchange via da and pa mechanisms we have shown that the frictional drag due to pa coupling is dominating . assuming infinite lifetime of the intermediate virtual phonon states , the frictional drag does not depend on the interlayer separation . according to our calculations the temperature dependence of the drag scattering rate divided by @xmath0 exhibits a pronounced peak both at matched electron density of two layers and at temperatures where transition from small to large angle scattering occurs . these results are in agreement with the finding of j. gramila , et al . @xcite . we acknowledge support of this work by the deutsche forschungsgemeinshaft ( grk : `` komplexitt in festkrpern '' ) . smb acknowledges also support from the deutscher akademischer austauschdienst ( daad ) and the us civilian research & development foundation for the independent states of the former soviet union ( crdf , award no . 375100 ) . for a review , see j. p. eisenstein , in perspectives in quantum hall effects : novel quantum liquids in low - dimensional semiconductor structures , edited by s. das sarma and a. pinczuk ( wiley , new - york , 1997 ) . e. lozovik and v. i. yudson , pisma zh . * 22 , * 556 ( 1975 ) [ jetp lett.22 , 271 ( 1975 ) ] ; zh . * 71 , * 738 ( 1976 ) [ sov . jetp * 44 , * 389 ( 1976 ) ] . s. i. shevchenko , fiz . nizk . * 2 , * 505 ( 1976 ) [ sov . j. low temp . phys.2 , 251 ( 1976 ) ] . y. w. suen , l. w. engel , m. b. santos , m. shayegan , and d. c. tsui , phys . rev * 68 , * 1379 ( 1992 ) . j. p. eisenstein , g. s. boebinger , l. n. pffeifer , k. w. west , and song he , phys . lett . * 68 , * 1383 ( 1992 ) . m. b. pogrebinskii , fiz . . poluprovodn . * 11 , * 637 ( 1977 ) [ sov . * 11 , * 372 ( 1977 ) ] . p. j. price , physica * 117 & 118 , * 750 ( 1983 ) . p. m. solomon , p. j. price , d. j. franck , and d. c. la tulipe , phys . * 63 , * 2508 ( 1989 ) . b. laikhtmann and p. m. solomon , phys . b * 41 , * 9921 ( 1990 ) . u. sivan , p. m. solomon , and h. shtrikman , phys . * 68 , * 1196 ( 1992 ) . t. j. gramila , j. p. eisenstein , a. h. macdonald , l. n. pffeifer , and k. w. west , phys . lett . * 66 , * 1216 ( 199 ) . t. j. gramila , j. p. eisenstein , a. h. macdonald , l. n. pffeifer , and k. w. west , surf . * 263 , * 446 ( 1992 ) . t. j. gramila , j. p. eisenstein , a. h. macdonald , l. n. pffeifer , and k. w. west , phys . b * 47 , * 12957 ( 1993 ) . h. rubel , e. h. linfield , d. a. ritchie , k. m. brown , m. pepper , and g. a. c. jones , semicond . * 10 , * 1229 ( 1995 ) . h. rubel , e. h. linfield , d. a. ritchie , k. m. brown , m. pepper , and g. a. c. jones , surf . * 361 & 362 , * 134 ( 1996 ) . m. c. bnsager , k. flensberg , b. y .- k . hu , and a .- p . jauho , phys . * 77 , * 1366 ( 1996 ) . n. p. r. hill , j. t. nicholls , e. h. linfield , m. pepper , d. a. ritchie , a. r. hamilton , and g. a. c. jones , j. phys . * 8 , * l557 ( 1996 ) . h. rubel , a. fischer , w. dietsche , k. von klitzing , and k. eberl , phys . * 78 , * 1763 ( 1997 ) . d. l. maslov , phys . b * 45 , * 1911 ( 1992 ) . l. zheng and a. h. macdonald , phys . b * 48 , * 8203 ( 1993 ) . e. lozovik and v. nikitkov , zh . * 111 , * 1107 ( 1997 ) [ sov . jetp * 84 , * 612 ( 1997 ) ] . h. c. tso , p. vasilopoulos , and f. m. peeters , phys . * 70 , * 2146 ( 1993 ) . jauho and h. smith , phys . b * 47 , * 4420 ( 1993 ) . h. c. tso , p. vasilopoulos , and f. m. peeters , phys . * 68 , * 2516 ( 1992 ) . v. f. gantmakher and y. b. levinson , scattering of carriers in metals and semiconductors ( noeth - holland , amsterdam , 1987 ) . m. kogan , fiz . tela * 4 , * 2474 ( 1962 ) [ sov . phys . solid state * 4 , * 1813 ( 1962 ) ] . j. d. zook , phys . rev . * 136a , * 869 ( 1964 ) . m. abramowitz and i. a. stegun , handbook of mathematical functions , dover publications inc . , new - york , 1972 m. a. chin , v. narayanamurti , h. l. stormer , and j. c. m. hwang , in phonon scattering in condensed matter , p. 141 , eds . w. eisenmenger , k. lamann , and s. dttinger , springer - verlag , berlin , 1984 .	we have calculated the temperature dependence of the frictional drag between spatially separated quantum wells with parallel two - dimensional electron gases due to interlayer electron - electron interaction mediated by virtual exchange of acoustic phonons due to piezoelectric and deformation potential interaction . it is shown that the frictional drag is dominated by the piezoelectric coupling . according to our calculations the temperature dependence of the drag scattering rate divided by @xmath0 exhibits a pronounced peak which for the experimental situation and in agreement with the finding of t. j. gramila , et al . phys . rev . b * 47 * , ( 1993 ) 12957 is obtained at about @xmath1 k. we ascribe the appearance of this peak to a change from small to large angle scattering in the virtual phonon exchange .
over a 7manifold , a topological reduction to a principal @xmath0bundle is achieved by the existence of a certain 3form . the fact that this 3form is generic or ( following the language of hitchin ) enables one to set up a variational principle over a fixed cohomology class whose critical points are precisely the manifolds of holonomy @xmath0 @xcite . in this paper we are concerned with a new type of riemannian geometry over 7manifolds which generalises this notion . topologically speaking , it is defined by an even or odd form which we think of as a _ spinor _ for the orthogonal bundle @xmath2 with its natural inner product of split signature . this construction is perfectly general and works in all dimensions , but there are special cases where @xmath3 acts with an _ open _ orbit in its spin representations @xmath4 . an example in dimension 6 are the so called generalised calabi yau manifolds associated with an @xmath5invariant spinor @xcite . in dimension 7 , there are stable spinors whose stabiliser is either conjugate to @xmath6 or @xmath7 . in this paper , we shall deal with the latter case and define a generalised @xmath0structure as a ( topological ) reduction from @xmath8 to @xmath7 . stability allows us to consider a generalised variational problem which provides us with various integrability conditions . we begin by introducing the algebraic setup . a reduction to @xmath7 gives rise to various objects . firstly , it induces a metric on @xmath1 , and moreover , a 2form @xmath9 which we refer to as the _ @xmath10field_. as an element of the lie algebra @xmath11 inside @xmath12 , it acts on any @xmath13representation by exponentiation . secondly , we obtain two unit spinors @xmath14 and @xmath15 in the irreducible spin representation of @xmath16 . tensoring the spinors yields an even or odd form @xmath17^{ev , od}$ ] by projection on @xmath18 . the first important result we prove states that any @xmath7invariant spinor @xmath19 can be expressed as @xmath20^{ev , od}.\ ] ] in physicists terminology , the scalar @xmath21 represents the _ dilaton _ here it appears as a scaling factor . moving on to global issues , we see that up to a @xmath10field transformation , a generalised @xmath0structure is essentially a pair of principle @xmath0fibre bundles inside the orthonormal frame bundle determined by the metric . in particular , any topological @xmath0manifold trivially induces a generalised @xmath0manifold , and consequently , any spinnable seven fold carries such a structure . over a compact manifold , we can classify generalised @xmath0structures up to vertical homotopy by an integer which effectively counts ( with an appropriate sign convention ) the number of points where the two @xmath0structures inside the @xmath16principal bundle coincide . over a closed manifold , we can then set up a variational problem along the lines of @xcite . in close analogy to the classical case , the condition for a critical point is that both the even and the odd @xmath7invariant spinor , now regarded as a form , have to be closed . we shall adopt this as the condition of _ strong _ integrability for any ( not necessarily compact ) generalised @xmath0structure . interpreting this integrability condition in terms of the right hand side of ( [ genformrep ] ) leads to our main result . theorem [ integrability ] characterises strongly integrable generalised @xmath0structures in terms of two linear metric connections @xmath22 with skew symmetric , closed torsion @xmath23 such that @xmath24 holds . connections with skew symmetric torsion have gained a lot of attention in the recent mathematical literature ( see , for instance , @xcite , @xcite , @xcite and @xcite ) due to their importance in string theory . eventually , our reformulation yields the supersymmetry equations arising in type ii supergravity with bosonic background fields @xcite . the spinorial picture is also useful for deriving geometrical properties . in particular , we compute the ricci tensor which is given by @xmath25 with @xmath26 denoting the hessian of the dilaton . a further striking consequence is a no go theorem for compact manifolds ( generalising similar statements in @xcite and @xcite ) : here , the torsion of any strongly integrable generalised @xmath0structure must vanish , that is , the underlying spinors @xmath14 and @xmath15 are parallel with respect to the levi civita connection . in this sense , only classical " solutions can be found for the variational problem . however , local examples with non vanishing torsion exist in abundance . using again the form definition of a generalised @xmath7structure , we will describe a systematic construction method known in string theory as @xmath1duality . it consists of changing the topology by replacing a fibre isomorphic to @xmath27 ( or more generally to an @xmath28torus ) without destroying integrability . this allows us to pass from a trivial generalised structure coming from a classical @xmath27invariant @xmath0structure with vanishing torsion ( for which we can easily find examples ) to a trivial @xmath27bundle with a non trivial generalised @xmath0structure . the lack of interesting compact examples motivated us to consider a _ constrained _ variational problem following ideas in @xcite . this gives rise to _ weakly _ integrable structures of either _ even _ or _ odd _ type . for these the no go theorem does not apply , but the construction of examples , let alone compact ones , remains an open problem . although arising out of a similar constraint as manifolds of weak holonomy @xmath0 , this notion gives rise to a new geometry without a classical counterpart which renders a straightforward application of @xmath1duality impossible . we investigate its properties along with those of the strongly integrable case . the algebraic theory outlined above not only makes sense for @xmath29structures , but also for @xmath30 inside @xmath31 , leading to the notion of a _ generalised @xmath16manifold_. it is also defined by an invariant form of mixed degree , which , however , is not stable . comparison with classical @xmath16geometry suggests closeness of this form as a natural notion of integrability . section [ spin7theory ] briefly explores the theory of these structures which is developed in full detail in @xcite . this paper is based on a part of the author s doctoral thesis @xcite . he wishes to acknowledge the daad , the studienstiftung des deutschen volkes , the epsrc and the university of oxford for various financial support . the author also wants to thank gil cavalcanti , michael crabb , claus jeschek , wilson sutherland and the examiners dominic joyce and simon salamon for helpful comments and discussions . finally , he wishes to express his special gratitude to his supervisor nigel hitchin . we consider the vector bundle @xmath32 , where @xmath1 is a real , @xmath28dimensional vector space . it carries a natural orientation and the inner product of signature @xmath33 , defined for @xmath34 and @xmath35 by @xmath36 singles out a group conjugate to @xmath37 inside @xmath38 . note that @xmath39 . as a @xmath40space , the lie algebra of @xmath37 decomposes as @xmath41 in particular , any 2form @xmath9 defines an element in the lie algebra @xmath42 . we will refer to such a 2form as a _ @xmath10field_. exponentiated to @xmath37 , its action on @xmath32 is @xmath43 next we define an action of @xmath32 on @xmath44 by @xmath45 as this squares to minus the identity it gives rise to an isomorphism @xmath46 . the exterior algebra @xmath47 becomes thus the pinor representation space of @xmath48 and splits into the irreducible spin representation spaces @xmath49 of @xmath50 . there is a canonical embedding @xmath51 of the identity component of @xmath40 into the spin group of @xmath32 . as a @xmath52module we have @xmath53 in analogy to the complex case . there @xmath54 , and the even and odd forms get twisted with the square root of the canonical line bundle . as long as we are doing linear algebra this is a mere notational issue but in the global situation we can not trivialise @xmath55 unless the manifold is orientable . in fact , a more refined analysis reveals that we can always choose a spin structure for @xmath32 whether the manifold is orientable or not such that the spinor bundle is isomorphic , as a @xmath40space , to the exterior form bundle , albeit in a non canonical way ( see , for instance , the discussion in section 2.8 in @xcite ) . we will neglect these subtleties for we will consider orientable manifolds only , and therefore omit the twist to ease notation . however it is important to bear it in mind when we set up the variational formalism in section [ variation ] . let @xmath56 be the clifford algebra anti automorphism defined on any element of degree @xmath57 by @xmath58 where @xmath59 for @xmath60 and @xmath61 for @xmath62 . the bilinear form @xmath63 where the subscript @xmath28 indicates taking the top degree component , is non degenerate and invariant under the action of @xmath50 . it is symmetric if @xmath64 and skew if @xmath65 . moreover , @xmath66 and @xmath67 are non degenerate and orthogonal if @xmath28 is even and totally isotropic if @xmath28 is odd . finally , we note that the action of a @xmath10field @xmath9 on a spinor @xmath68 is given by @xmath69 in this paper we shall be concerned with special structures on @xmath32 , which we describe in terms of reductions to special subgroups of @xmath70 or @xmath71 . a _ generalised metric structure _ is a reduction from @xmath70 to @xmath72 . figure [ lightcone ] suggests how to characterise a metric splitting algebraically . if we think of the coordinate axes @xmath1 and @xmath73 as a lightcone , choosing a subgroup conjugate to @xmath72 inside @xmath70 boils down to the choice of a spacelike @xmath74 and a timelike orthogonal complement @xmath75 . interpreting @xmath74 as the graph of a linear map @xmath76 yields a metric @xmath77 and a 2form @xmath9 as the symmetric and the skew part of the corresponding bilinear form @xmath78 . indeed we have @xmath79 so that @xmath77 is positive definite . as @xmath74 and @xmath75 are orthogonal , taking @xmath75 instead of @xmath74 yields the same 2form @xmath9 but the metric @xmath80 . conversely , assume we are given a metric @xmath77 and a 2form @xmath9 on @xmath1 . if we transform the diagonal @xmath81 by @xmath82 , we obtain a splitting @xmath83 inducing @xmath77 and @xmath9 . [ genmetricchar ] the choice of an equivalence class in the space @xmath84 is equivalent to either set of the following data : \(i ) a metric splitting @xmath85 into subbundles @xmath86 and @xmath87 with positive and negative definite metrics @xmath88 . \(ii ) a riemannian metric @xmath77 and a 2form @xmath9 on @xmath1 . note that a reduction to @xmath72 determines @xmath77 and @xmath9 up to a common scalar which , however , is fixed by the explicit choice of the inner product ( [ innprod ] ) . @xmath89 in the same vein , we call an element of @xmath90 a _ generalised oriented metric structure _ , which corresponds to a metric @xmath77 , a @xmath10field @xmath9 and an orientation on @xmath1 . since the bundle @xmath32 is always spinnable , we can also lift the discussion to the group @xmath50 . moreover , in some situations it is natural to introduce an additional scalar , that is , we enhance the structure group @xmath50 to the conformal spin group @xmath3 so that a reduction to @xmath91 gives a further degree of freedom . this provides the right framework within which we can discuss generalised @xmath0structures , to which we turn next . a _ generalised @xmath0structure _ is a reduction from the structure group @xmath8 of @xmath32 to @xmath7 . we want to characterise generalised @xmath0structures along the lines of proposition [ genmetricchar ] . to get things rolling , we first look at the tensorial invariants on @xmath1 which are induced by such a reduction . since @xmath7 determines some group @xmath92 conjugate to @xmath93 , it induces a generalised oriented metric structure @xmath94 . the isomorphisms @xmath95 allow us to transport the respective @xmath0structure on @xmath74 and @xmath75 to the tangent bundle . the resulting @xmath0structures @xmath96 and @xmath97 inside @xmath98 give rise to two unit spinors @xmath14 and @xmath15 in the irreducible spin representation @xmath99 of @xmath100 . on the other hand , @xmath7 also acts on @xmath44 as a subgroup of @xmath13 . to relate these two actions we consider the following construction which is basically the classical identification of @xmath101 with @xmath102 followed by a twist with @xmath103 . we write the clifford algebra @xmath48 as a @xmath104graded tensor product @xmath105 where the isomorphism is given by extension of the map @xmath106 . the maps @xmath107 induce isomorphisms @xmath108 mapping @xmath109 isomorphically onto @xmath110 . consequently , the compounded algebra isomorphism @xmath111 maps @xmath30 onto @xmath112 inside @xmath13 . let @xmath113 denote the @xmath16invariant inner product on @xmath114 . for two spinors @xmath14 and @xmath15 the _ pinor product _ @xmath115 is the endomorphism of @xmath114 defined by @xmath116 on the other hand , there exists an algebra isomorphism @xmath117 . projection on the first summand induces a matrix representation for clifford multiplication . for concrete computations , we realise this representation by @xmath118 where @xmath119 , @xmath120 is the standard basis of skew symmetric endomorphisms of @xmath114 , taken with respect to an orthonormal basis @xmath121 . this relates to the pinor product by @xmath122 let @xmath123 denote the canonical vector space isomorphism between @xmath124 and @xmath44 . we consider the map @xmath125:\delta\otimes\delta\stackrel{{\hat{\circ}}\oplus0}{\longrightarrow } \end(\delta)\oplus\end(\delta)\stackrel{\kappa^{-1}}{\longrightarrow}\cliff(t , g)\stackrel{j}{\longrightarrow } \lambda^t^\ ] ] and think of any element in @xmath101 as a form . we denote by @xmath17^{ev , od}$ ] the projection on the even or odd part and add a subscript @xmath9 if we wedge with the exponential @xmath103 . the following result states that up to a sign twist , the action of @xmath1 on @xmath101 and @xmath18 commute . [ lmap ] for any @xmath126 and @xmath127 we have @xmath128_b^{ev , od } & = & \phantom{\pm}\iota_b(x{\hat{\otimes}}1)\bullet \,[\psi_+\otimes\psi_-]_b^{od , ev}\nonumber\\ \,[\psi_+\otimes x\cdot\psi_-]_b^{ev , od } & = & \pm\iota_b(1{\hat{\otimes}}x)\bullet [ \psi_+\otimes\psi_-]_b^{od , ev}.\nonumber\end{aligned}\ ] ] in particular , the forms @xmath17^{ev , od}_b$ ] are @xmath7invariant . first we assume that @xmath129 . by convention , we let @xmath1 act through the inclusion @xmath130 followed by projection on the first summand . thus @xmath128 & = & j\big(\kappa^{-1}(x\cdot\psi_+{\hat{\circ}}\psi_-\oplus0)\big)\\ & = & j\big(x\cdot \kappa^{-1}(\psi_+{\hat{\circ}}\psi_-)\big)\\ & = & -x\llcorner [ \psi_+\otimes\psi_-]+x\wedge [ \psi_+\otimes\psi_-]\\ & = & \iota_0(x{\hat{\otimes}}1)\bullet [ \psi_+\otimes\psi_-],\end{aligned}\ ] ] where we have used ( [ cliffordid ] ) and the identity @xmath131 which holds for any clifford algebra @xmath132 and @xmath133 . we can argue similarly for @xmath134 and we obtain @xmath135^{ev}=\iota_0(x{\hat{\otimes}}1)\bullet [ \psi_+\otimes\psi_-]^{od},\quad [ x\cdot\psi_+\otimes\psi_-]^{od}=\iota_0(x{\hat{\otimes}}1)\bullet [ \psi_+\otimes\psi_-]^{ev},\ ] ] so that the case @xmath136 is shown . now let @xmath9 be an arbitrary @xmath10field . for the sake of clarity we will temporarily denote by @xmath137 the exponential map from @xmath12 to @xmath13 , while the untilded exponential takes values in @xmath138 . the adjoint representation @xmath139 of the group of units inside a clifford algebra @xmath132 restricts to the double cover @xmath140 still denoted by @xmath139 . as a transformation in @xmath141 we then have @xmath142 . since @xmath143)=4v\wedge w$ ] and the vectors @xmath144 are isotropic with respect to the inner product @xmath145 , in our situation we get @xmath146 and thus @xmath147 for the @xmath10field @xmath148 . hence @xmath149^{ev , od}_b & = & \big(\widetilde{e}^{\:b/2}\bullet\iota_0(x{\hat{\otimes}}1\big)\bullet\widetilde{e}^{\:-b/2})\bullet [ \psi_+\otimes\psi_-]^{od , ev}_b\\ & = & ad(\widetilde{e}^{b/2})\big(\iota_0(x{\hat{\otimes}}1)\big)\bullet [ \psi_+\otimes\psi_-]^{od , ev}_b\\ & = & \iota_b(x{\hat{\otimes}}1)\bullet [ \psi_+\otimes\psi_-]^{od , ev}_b.\end{aligned}\ ] ] similarly , the claim is checked for @xmath150^{ev , od}_b$ ] which completes the proof . the identification @xmath151 also enables us to derive a normal form for @xmath17^{ev , od}$ ] by choosing a suitable orthonormal frame . the coefficients of the homogeneous components are given up to a scalar by @xmath152 . since the action of @xmath16 on the stiefel variety @xmath153 , the set of pairs of orthonormal spinors , is transitive , we may assume that @xmath154 and @xmath155 , where @xmath156 and @xmath157 is shorthand for @xmath158 and @xmath159 with @xmath160 . using the representation ( [ matrep ] ) , the computation of the normal form is straightforward . [ normalform ] there exists an orthonormal basis @xmath161 such that @xmath162^{ev } & = & \phantom{+}\cos(a)+\sin(a)(-e_{23}-e_{45}+e_{67})+\\ & & + \cos(a)(-e_{1247}+e_{1256}+e_{1346}+e_{1357}-e_{2345}+e_{2367}+e_{4567})+\\ & & + \sin(a)(e_{1246}+e_{1257}+e_{1347}-e_{1356})-\sin(a)e_{234567}\end{aligned}\ ] ] and @xmath162^{od } & = & \phantom{+}\sin(a)e_1+\sin(a)(e_{247}-e_{256}-e_{346}-e_{357})+\\ & & + \cos(a)(e_{123}+e_{145}-e_{167}+e_{246}+e_{257}+e_{347}-e_{356})+\\ & & + \sin(a)(-e_{12345}+e_{12367}+e_{14567})+\cos(\alpha)e_{1234567}.\end{aligned}\ ] ] if the spinors @xmath14 and @xmath15 are linearly independent , we can express the structure form in terms of the invariants associated with @xmath163 , the stabiliser of the pair @xmath164 . [ normalformcor ] if @xmath165 denotes the dual of the unit vector in @xmath1 which is stabilised by @xmath163 , @xmath166 the khler form and @xmath167 the real and imaginary parts of the holomorphic volume form , then @xmath168^{ev}=c+s\omega+c(\alpha\wedge\psi_--\frac{1}{2}\omega^2)-s\alpha\wedge\psi_+-\frac{1}{6}s\omega^3\ ] ] and @xmath168^{od}=s\alpha - c(\psi_++\alpha\wedge\omega)-s\psi_--\frac{1}{2}s\alpha\wedge\omega^2+cvol_g.\ ] ] moreover , @xmath169 . if @xmath170 , then both @xmath0structures coincide and @xmath168^{ev}=1-\star\varphi,\quad[\psi_+\otimes\psi_-]^{od}=-\varphi+vol_g,\ ] ] where @xmath171 is the stable 3form associated with @xmath0 . the stabiliser of the forms @xmath172 and @xmath173 is isomorphic to @xmath6 @xcite . the @xmath174 degrees of freedom which parametrise a reduction from @xmath13 to @xmath7 are exhausted by @xmath175 degrees of freedom for the choice of a metric @xmath77 , @xmath176 for a 2form @xmath9 and twice @xmath177 degrees for two unit spinors @xmath14 and @xmath15 . however , this data does not achieve a full description of @xmath7invariant spinors yet as these are _ stable _ , a notion due to hitchin @xcite . a spinor @xmath19 in @xmath178 is said to be _ stable _ if @xmath19 lies in an open orbit under the action of @xmath3 . in @xcite sato and kimura classified the representations of complex reductive lie groups which admit an open orbit . apart from dimension 7 , stable spinors in the sense of the definition above only occur in dimension 6 and give rise to _ generalised calabi structures _ associated with the group @xmath5 @xcite . the key point here is that in both cases we obtain an invariant volume form . in our case , this generalises the concept of stability for a @xmath0invariant 3form @xmath171 with associated volume form @xmath179 @xcite . to begin with , we note that the spin representation @xmath114 of @xmath16 is real , and so is the tensor product @xmath101 . consequently , there is ( up to a scalar ) a unique invariant in @xmath180 , or equivalently , @xmath30equivariant maps @xmath181 . morally these are given by the hodge @xmath182operator twisted with the @xmath10field and the anti . the _ box operator _ or _ generalised hodge @xmath182operator _ @xmath183 associated with the generalised metric @xmath94 is defined by @xmath184 if @xmath77 and @xmath9 are induced by @xmath19 we will also use the sloppier notation @xmath185 or drop the subscript altogether . for present and later use , we note the following lemma whose proof is immediate from the definitions . [ hat ] let @xmath186 . then @xmath187 and @xmath188 . [ selfdual ] for any @xmath189 @xmath190_b=[\psi_+\otimes\psi_-]_b\ ] ] or equivalently , @xmath190_b^{ev , od}=[\psi_+\otimes\psi_-]_b^{od , ev}.\ ] ] according to lemma [ hat ] , @xmath191_b & = & e^{b/2}\wedge\star\sigma[\psi_+\otimes\psi_-]\\ & = & e^{b/2}\wedge j\big(\kappa^{-1}(\psi_+{\hat{\circ}}\psi_-)\cdot vol\big)\end{aligned}\ ] ] where we used the general identity @xmath192 . but @xmath193 is just @xmath194 , whence the assertion . let @xmath195 denote the space of @xmath7invariant spinors in @xmath196 or @xmath197 . with any element of @xmath195 we associate a volume form @xmath198 now the @xmath199operator transforms naturally under the lift @xmath200 of any element @xmath201 which means that @xmath202 therefore , we immediately conclude [ volformg2g2 ] @xmath203 is homogeneous of degree 2 and @xmath13invariant . an explicit coordinate description of the complexified invariant was given in @xcite . however , with the aim of setting up the variational principle , this formulation proved to be rather cumbersome for our purposes , which motivated our approach in terms of @xmath204structures . we will also need the differential of this map . since the form @xmath145 is non degenerate , we can write @xmath205 for a unique @xmath206 , the _ companion _ of @xmath19 , which is also a @xmath7invariant spinor . by rescaling @xmath203 appropriately , we conclude that @xmath207 . from @xmath203 we derive a further invariant attached to a @xmath29invariant spinor @xmath19 , namely a real scalar @xmath208 which we refer to as the _ dilaton_. it is defined by @xmath209 using the normal form description of proposition [ normalform ] we obtain @xmath210^{ev , od}_b\big)=8vol_g$ ] , hence @xmath211^{ev , od}_b$ ] . we summarise the results of this section in the following [ representation ] generalised @xmath0structures are in 11 correspondence with lines of spinors @xmath19 in @xmath212 ( or @xmath213 ) whose stabiliser under the action of @xmath13 is isomorphic to @xmath7 . we refer to @xmath19 as the _ structure form _ of the generalised @xmath0structure . this form can be uniquely written ( modulo a simultaneous sign change for @xmath14 and @xmath15 ) as @xmath214^{ev}_b\ ] ] for a 2form @xmath9 , two unit spinors @xmath215 and a real scalar @xmath21 . in particular , the space of @xmath7invariant spinors is open . a _ topological generalised @xmath0structure _ over a 7manifold @xmath216 is a topological @xmath7reduction of the @xmath8principal bundle associated with @xmath32 . it is characterised by a stable even or odd spinor @xmath19 which we view as a form . consequently , we will denote this structure by the pair @xmath217 and call @xmath19 the _ structure form_. we will usually drop the adjective topological " and simply refer to a generalised @xmath0structure . as we saw earlier , a generalised @xmath0structure induces a generalised metric . in particular there exists a metric @xmath77 or equivalently an @xmath218principal fibre bundle which admits two @xmath0subbundles . the inclusion @xmath219 implies that the underlying manifold is spinnable and distinguishes a preferred spin structure for which we consider the associated spinor bundle @xmath114 . using theorem [ representation ] we can now assert the following statement . [ topology ] a topological generalised @xmath0structure @xmath217 is characterised by the following data : * an orientation * a metric @xmath77 * a 2form @xmath9 * a scalar function @xmath21 * two unit spinors @xmath220 such that @xmath221_b=\rho+\box_{g , b}\rho.\ ] ] a trivial example of a generalised @xmath0structure is a topological @xmath0manifold with associated spinor @xmath222 . a generalised structure arising in this way ( possibly with a non trivial @xmath10field and dilaton ) is said to be _ straight_. the existence of a nowhere vanishing spinor field in dimension 7 is a classical result @xcite and implies a 7fold @xmath216 carries a topological generalised @xmath0structure if and only if @xmath216 is spinnable . next we discard the @xmath10field and the dilaton and focus on the @xmath0structures induced by @xmath14 and @xmath15 . our aim is to classify the @xmath0structures up to equivalence under a @xmath16fibre bundle isomorphism . since the classification of principal fibre bundles is a problem of homotopy theory , this boils down to deform homotopically @xmath14 into @xmath15 through sections . more concretely , we regard @xmath0structures as being defined by ( continuous ) sections of the sphere bundle @xmath223 associated with @xmath114 . on the space of sections @xmath224 we introduce the following equivalence relation . two spinors @xmath14 and @xmath15 are considered to be equivalent ( denoted @xmath225 ) if and only if there exists a continuous map @xmath226 such that @xmath227 , @xmath228 and @xmath229 . an equivalence class will be denoted by @xmath230 $ ] . if two sections are vertically homotopic , then the corresponding @xmath0structures are isomorphic as principal @xmath0bundles over @xmath216 . in particular the generalised structure defined by the pair @xmath231 is equivalent to a straight structure if and only if @xmath225 . what we aim to determine is the set of generalised structures with fixed @xmath14 , i.e. @xmath232 , the set of isomorphism classes of principal @xmath0fibre bundles . if a generalised structure is defined by two inequivalent spinors , then it is said to be _ exotic_. here is an example . consider @xmath233 . since the tangent bundle of @xmath234 is trivial , so is the sphere bundle of @xmath114 , i.e. @xmath235 . consequently , @xmath236=\pi_7(s^7)={\mathbb{z}}\ ] ] and any map @xmath237 which is not homotopic to a constant gives rise to an exotic structure for @xmath238 . in general , the question whether or not two sections are vertically homotopic can be tackled by using obstruction theory ( see for instance @xcite ) . assume that we are given a fibre bundle with connected fibre over a not necessarily compact @xmath28fold @xmath239 , and two sections @xmath240 and @xmath241 which are vertically homotopy equivalent over the @xmath113skeleton @xmath242}$ ] of @xmath216 . the obstruction for extending the vertical homotopy to the @xmath243skeleton lies in @xmath244 . in particular , there is the first non trivial obstruction @xmath245 for the least integer @xmath246 such that @xmath247 , called the _ primary difference _ of @xmath240 and @xmath241 . it is a homotopy invariant of @xmath240 and @xmath241 and enjoys the additivity property @xmath248 coming back to the generalised @xmath0case we consider the sphere bundle @xmath249 over @xmath250 with fibre @xmath234 . consequently , the primary difference of two sections lies in @xmath251 and this is the only obstruction for two sections to be vertically homotopy equivalent . the additivity property implies that @xmath252 if and only if @xmath253 , that is , @xmath254 . moreover , for any class @xmath255 there exists a section @xmath256 such that @xmath257 @xcite . as a consequence , we obtain the the set of generalised @xmath0structures can be identified with @xmath258 modulo @xmath259 , generalised @xmath0structures are therefore classified by an integer invariant which over a compact @xmath250 has the natural interpretation as the number of points ( counted with an appropriate sign convention ) where the two @xmath0structures coincide . to see this we associate with every equivalence class @xmath260 $ ] the intersection class @xmath261 of the 7dimensional oriented submanifolds @xmath262 and @xmath263 inside @xmath249 . since the total space of the sphere bundle is 14dimensional , the intersection class counts the number of points in @xmath216 where the two spinors @xmath14 and @xmath15 coincide . taking the cup product of the poincar duals of @xmath263 and @xmath262 sets up a map @xmath264\in gen(m)\mapsto pd\big(\psi_+(m)\big)\cup pd\big(\psi_-(m)\big)\in h^{14}(\mathbb{s},{\mathbb{z}}).\ ] ] on the other hand , integration along the fibre defines an isomorphism @xmath265 by the gysin sequence . therefore , any generalised @xmath0structure induced by the equivalence class @xmath260 $ ] over a compact 7manifold @xmath216 gives rise to a well defined cohomology class @xmath266 . the following theorem shows this class to coincide with the primary difference . for the proof , i benefited from discussions with w. sutherland and m. crabb . [ intersection ] if @xmath216 is compact , then @xmath267 in particular generalised @xmath0structures are classified by the number of points where the two underlying @xmath0structures coincide . we regard the spinor bundle @xmath114 as an 8dimensional oriented real vector bundle over m and consider the two sections @xmath14 and @xmath15 of the sphere bundle . deform the section @xmath268 of the pullback of @xmath114 to @xmath269 to be transverse to the zero section . the primary difference @xmath270 can then be represented geometrically by the ( signed ) zero set of this deformation . in particular , if @xmath14 and @xmath271 never coincide , then the primary difference is 0 . therefore the intersection number , defined geometrically by making @xmath262 and @xmath263 transverse and taking the coincidence set , will be @xmath272 ( with appropriate sign conventions ) . by virtue of ( [ additivity ] ) , we have @xmath273 . the difference class @xmath274 corresponds to the self @xmath275 which is 0 since @xmath216 is 7dimensional , whence the assertion . assume @xmath216 to be a closed and oriented manifold which carries a topological generalised @xmath0structure defined by the stable spinor @xmath19 . we shall set up a variational problem along the lines of @xcite and @xcite . from a @xmath276point of view , @xmath19 is a section of the vector bundle whose fibre is @xmath277 ( cf . the remark in section [ generalisedmetric ] ) . untwisting by the line bundle @xmath278 , we obtain a corresponding open set still denoted by @xmath195 , on which we can consider the induced volume functional @xmath203 as described in proposition [ volformg2g2 ] . it is homogeneous of degree 2 and therefore , it defines a @xmath279equivariant function @xmath280 since @xmath281 associated with the @xmath276principal fibre bundle over @xmath216 , @xmath203 thus takes values in @xmath282 and we obtain the volume functional @xmath283 defined over stable forms . as stability is an open condition , we can differentiate this functional and consider its variation over a fixed cohomology class . instead of working with ordinary cohomology only we will allow for an extra twist by a closed 3form @xmath284 . this means that we replace the differential operator @xmath285 by the twisted operator @xmath286 . closeness of @xmath284 guarantees that @xmath287 still defines a differential complex . moreover , we will also consider the following constraint . the bilinear form @xmath145 induces a non degenerate pairing between @xmath288 and @xmath289 defined by integration of @xmath290 over @xmath216 . if @xmath291 is @xmath284exact , stokes theorem and the definition of the involution @xmath56 imply that @xmath292 this vanishes for all @xmath293 if and only if @xmath68 is @xmath284closed . consequently , we can identify @xmath294 the exterior differential @xmath287 maps the latter space isomorphically onto @xmath295 so that @xmath296 finally , we obtain the non degenerate quadratic form on @xmath295 given by @xmath297 the same conclusion holds for odd instead of even forms . let @xmath284 be a closed 3form . \(i ) a @xmath287closed stable form @xmath298 is a critical point in its cohomology class if and only if @xmath299 . \(ii ) a @xmath287exact form @xmath298 defines a critical point subject to the constraint @xmath300 if and only if there exists a real constant @xmath301 with @xmath302 . the first variation of @xmath303 is @xmath304 to find the unconstrained critical points we have to vary over a fixed @xmath287cohomology class , i.e. @xmath305 . as we saw in ( [ stokes ] ) , we have @xmath306 and this vanishes for all @xmath293 if and only if @xmath307 is @xmath287closed . in the constrained case the differential of @xmath308 at @xmath19 is @xmath309 by lagrange s theorem , we see that for a critical point we need @xmath310 . we adopt these various conditions for defining a critical point as integrability condition of a topological generalised @xmath0structure . [ intstruc ] let @xmath284 be a closed 3form and @xmath301 be a real , non zero constant . \(i ) a topological generalised @xmath0structure @xmath217 is said to be _ strongly integrable _ with respect to @xmath284 if and only if @xmath311 \(ii ) a topological generalised @xmath0structure @xmath217 is said to be _ weakly integrable with respect to @xmath284 and with killing number @xmath301 _ if and only if @xmath312 we call such a structure _ even _ or _ odd _ according to the parity of the form @xmath19 . if we do not wish to distinguish the type , we will refer to both structures as _ weakly integrable_. similarly , the structures in ( i ) and ( ii ) will also be referred to as _ integrable _ if a condition applies to both weakly and strongly integrable structures . as we shall see in corollary [ torsioncomp ] , the number @xmath301 represents the @xmath313torsion form of the underlying @xmath0structures which is why we refer to it as the killing number @xcite . consider a straight topological @xmath0manifold @xmath314 equipped with an additional closed 3form @xmath284 . according to corollary [ normalformcor ] and theorem [ representation ] , the corresponding structure form is @xmath315 with companion @xmath316 ) . writing @xmath317 , we want to solve the equations of strong integrability @xmath318 it follows immediately that this is equivalent to @xmath319 , @xmath320 and @xmath321 , @xmath322 , that is , the holonomy is contained in @xmath0 . if we ask for weak integrability the question only makes sense in the even case as @xmath323 for structures of odd type . the equation of weak integrability becomes @xmath324 implying @xmath319 , @xmath325 and @xmath326 . since @xmath327 , we have @xmath328 . a straight structure can therefore never induce a weakly integrable structure in this sense , weak integrability has no classical counterpart . next we want to interpret the integrability conditions in definition [ intstruc ] in terms of the data of theorem [ topology ] . for a vanishing @xmath10field , the identification @xmath151 transforms the twisted dirac operators @xmath329 and @xmath330 on @xmath101 , given locally by @xmath331 into the dirac operators on @xmath57forms @xmath332 and @xmath333 @xcite . here and in the sequel , @xmath334 designates the levi civita connection on the tangent bundle @xmath1 or its lift to @xmath114 . the transformation under a non trivial @xmath10field is given in the next proposition . [ dirac ] we have @xmath335^{ev , od}_b & = & \phantom{+}d[\psi_+\otimes\psi_-]^{od , ev}_b+d^{\box}[\psi_+\otimes\psi_-]^{od , ev}_b\\ & & + \frac{1}{2}e^{b/2}\wedge\big(db\llcorner [ \psi_+\otimes\psi_-]^{od , ev}-db\wedge [ \psi_+\otimes\psi_-]^{od , ev}\big)\\ \,[\widehat{{\mathcal}{d}}(\psi_+\otimes\psi_-)]^{ev , od}_b & = & \pm d[\psi_+\otimes\psi_-]^{od , ev}_b\mp d^{\box}[\psi_+\otimes\psi_-]^{od , ev}_b\\ & & \mp\frac{1}{2}e^{b/2}\wedge\big(db\llcorner[\psi_+\otimes\psi_-]^{od , ev}+ db\wedge [ \psi_+\otimes\psi_-]^{od , ev}\big),\end{aligned}\ ] ] where @xmath336 . for @xmath136 this is just the classical case mentioned above . for an arbitrary @xmath10field @xmath9 we have @xmath337^{ev , od}_b = e^{b/2}\wedge [ { \mathcal}{d}(\psi_+\otimes\psi_-)]^{ev , od}=e^{b/2}\wedge d[\psi_+\otimes\psi_-]^{od , ev}+e^{b/2}\wedge d^{\star}[\psi_+\otimes\psi_-]^{od , ev}.\ ] ] the first term on the right hand side equals @xmath338^{od , ev}=d[\psi_+\otimes\psi_-]^{od , ev}_b-\frac{1}{2}e^{b/2}\wedge db\wedge [ \psi_+\otimes\psi_-]^{od , ev}\ ] ] while for the second term , we have @xmath339^{od , ev}=\mp\!\star{d\!\star\!}[\psi_+\otimes\psi_-]^{od , ev}.\ ] ] proposition [ selfdual ] and lemma [ hat ] give @xmath340^{od , ev } & = & \mp e^{b/2}\wedge\star d\big(e^{b/2}\wedge\sigma[\psi_+\otimes\psi_-]^{ev , od}_b\big)\\ & = & \mp\frac{1}{2 } e^{b/2}\wedge\star \big(db\wedge \sigma[\psi_+\otimes\psi_-]^{ev , od}\big)\mp\box_{g , b}d[\psi_+\otimes\psi_-]^{ev , od}_b\\ & = & \phantom{\pm}\frac{1}{2 } e^{b/2}\wedge\big(db\llcorner [ \psi_+\otimes\psi_-]^{od , ev})+d^{\box}[\psi_+\otimes\psi_-]^{od , ev}_b,\end{aligned}\ ] ] hence @xmath341^{ev , od}_b & = & \phantom{+}d[\psi_+\otimes\psi_-]^{od , ev}_b+d^{\box}[\psi_+\otimes\psi_-]_b^{od , ev}\\ & & + \frac{1}{2}e^{b/2}\wedge\big(db\llcorner[\psi_+\otimes\psi_-]^{od , ev}-db\wedge [ \psi_+\otimes\psi_-]^{od , ev}\big).\end{aligned}\ ] ] similarly , we obtain @xmath342^{ev , od}_b & = & \pm e^{b/2}\wedge\big(d[\psi_+\otimes\psi_-]^{od , ev}\pm\star { d\!\star\!}[\psi_+\otimes\psi_-]^{od , ev}\big)\\ & = & \pm d[\psi_+\otimes\psi_-]_b^{od , ev}\mp d^{\box}[\psi_+\otimes\psi_-]^{od , ev}_b\\ & & \mp\frac{1}{2}e^{b/2}\wedge\big(db\llcorner [ \psi_+\otimes\psi_-]^{od , ev}+ db\wedge [ \psi_+\otimes\psi_-]^{od , ev}\big).\end{aligned}\ ] ] we now come to our main result . [ integrability ] a generalised @xmath0structure @xmath217 is weakly integrable with respect to @xmath284 and killing number @xmath301 if and only if @xmath343_b=\rho+\box_{g , b}\rho$ ] satisfies ( with @xmath317 ) @xmath344 and @xmath345 in case of an even structure and @xmath346 in case of an odd structure . the structure @xmath343_b=\rho+\box_{g , b}\rho$ ] is strongly integrable if and only if these equations hold for @xmath328 . we refer to the equation involving the covariant derivative of the spinor as the _ generalised killing equation _ and to the equation involving the differential of @xmath21 as the _ dilatino equation_. the generalised killing equation basically states that we have two metric connections @xmath22 preserving the underlying @xmath347structures whose torsion ( as it is to be defined in the next section ) is _ skew symmetric_. the dilatino equation then serves to identify the components of the torsion with respect to the decomposition into irreducible @xmath347modules with the additional data @xmath348 and @xmath301 . the generalised killing and the dilatino equation occur in physics as solutions to the supersymmetry variations in type ii superstring theory with bosonic background fields @xcite . assume that @xmath19 is even and satisfies @xmath349 . the odd case is dealt with in a similar fashion . applying the @xmath199operator and using @xmath350 , we obtain @xmath351^{od}_b = e^{b/2}\wedge \big(h\llcorner [ e^{-\phi}\psi_+\otimes\psi_-]^{od}+\lambda [ e^{-\phi}\psi_+\otimes\psi_-]^{ev}\big).\ ] ] consequently , proposition [ dirac ] implies @xmath352^{ev } = t\llcorner [ e^{-\phi}\psi_+\otimes\psi_-]^{od}- t\wedge [ e^{-\phi}\psi_+\otimes\psi_-]^{od}+\lambda [ e^{-\phi}\psi_+\otimes\psi_-]^{ev}.\ ] ] as a corollary to proposition [ lmap ] , we see that @xmath353^{ev , od } = \frac{1}{8}\big(-t\cdot\psi_+\otimes\psi_-\pm\psi_+\otimes t\cdot\psi_-\mp \sum\limits_i(e_i\llcorner t)\cdot\psi_+\otimes e_i\cdot\psi_-+\sum\limits_ie_i\cdot\psi_+\otimes[e_i\llcorner t)\cdot\psi_-]^{od , ev}\ ] ] and @xmath354^{ev , od } = \frac{1}{8}[t\cdot\psi_+\otimes\psi_-\pm\psi_+\otimes t\cdot\psi_-\mp \sum\limits_i(e_i\llcorner t)\cdot\psi_+\otimes e_i\cdot\psi_--\sum\limits_ie_i\cdot\psi_+\otimes(e_i\llcorner t)\cdot\psi_-]^{od , ev}.\ ] ] hence ( [ ldirac ] ) entails @xmath355^{ev}=\frac{1}{4}e^{-\phi}[- t\cdot\psi_+\otimes\psi_-+\sum\limits_ie_i\cdot\psi_+\otimes ( e_i\llcorner t)\cdot\psi_-+\lambda\psi_+\otimes\psi_-]^{ev}.\ ] ] let @xmath356 denote the dirac operator associated with the clifford bundle @xmath357 . contraction with @xmath358 yields @xmath359 and therefore @xmath360 from this expression we deduce @xmath361 since @xmath362 . it follows @xmath363 we derive the corresponding expression for the spinor @xmath14 by using @xmath364 which accounts for the minus sign . next we turn to the dilatino equation . from @xmath365 we deduce @xmath366 on the other hand , contracting ( [ inteq1 ] ) with @xmath367 yields @xmath368 putting the last two equations together results in the dilatino equation for @xmath14 . we can perform the same calculation with @xmath364 instead of @xmath329 to derive the dilatino equation for @xmath15 . conversely assume that the generalised killing and dilatino equations of an even structure hold the odd case , again , is analogous . we note that @xmath369^{ev}_b = e^{-\phi}\lambda[\psi_+\otimes\psi_-]^{od}_b\ ] ] is equivalent to @xmath370^{ev}-d\phi\wedge [ \psi_+\otimes\psi_-]^{ev}=\lambda [ \psi_+\otimes\psi_-]^{od}-t\wedge [ \psi_+\otimes\psi_-]^{ev}.\ ] ] now @xmath371^{ev}-d\phi\wedge [ \psi_+\otimes\psi_-]^{ev } & = & \phantom{+}\frac{1}{2}[-d\phi\cdot\psi_+\otimes\psi_-+\psi_+\otimes d\phi\cdot\psi_-]^{od}+\\ & & + \sum\limits_i e_i\wedge\nabla_{e_i}[\psi_+\otimes\psi_-]^{ev}\\ & = & \phantom{+}\lambda [ \psi_+\otimes\psi_-]^{od}+\frac{1}{4 } [ t\cdot\psi_+\otimes\psi_-+\psi_+\otimes t\cdot\psi_-]^{od}+\\ & & + \frac{1}{2}\sum\limits_i[e_i\cdot\nabla_{e_i}\psi_+\otimes\psi_-- \nabla_{e_i}\psi_+\otimes e_i\cdot\psi_-+\\ & & + e_i\cdot\psi_+\otimes\nabla_{e_i}\psi_--\psi_+\otimes e_i\cdot\nabla_{e_i}\psi_-]^{od}\\ & = & \phantom{+}\lambda [ \psi_+\otimes\psi_-]^{od}-\frac{1}{8 } [ t\cdot\psi_+\otimes\psi_-+\psi_+\otimes t\cdot\psi_-\\ & & -\sum\limits_i ( e_i\llcorner t)\cdot\psi_+\otimes e_i\cdot\psi_--\sum\limits_i e_i\cdot\psi_+\otimes(e_i\llcorner t)\cdot\psi_-]^{od}\\ & = & \phantom{-}\lambda [ \psi_+\otimes\psi_-]^{od } -t\wedge [ \psi_+\otimes\psi_-]^{ev}\end{aligned}\ ] ] which proves ( [ closed ] ) and thus the theorem . the theorem holds more generally for any 3form @xmath1 , closed or not . similarly , we can introduce a 1form @xmath165 and consider the twisted differential operator @xmath372 . this substitutes @xmath373 by @xmath374 in the dilatino equation . in the light of the spinorial formulation of integrability , we shall from now on always consider the twisted differential @xmath375 applied to a @xmath10field free form @xmath343^{ev , od}$ ] . we refer to the 3form @xmath1 as the _ torsion _ of the generalised structure . generally speaking , the _ torsion tensor _ @xcite is defined by @xmath376 and measures the difference between an arbitrary metric connection @xmath377 and the levi civita connection @xmath334 . in our situation the spinors @xmath14 and @xmath15 induce two @xmath0subbundles which carry metric connections @xmath378 and @xmath379 such that @xmath380 it therefore makes sense to consider the broader class of geometries defined by two linear metric connections @xmath22 with skew symmetric and closed torsion @xmath23 . this class encapsulates all the structures we obtain by applying the variational principle and conveniently avoids distinguishing between internal " torsion @xmath381 coming from the ubiquitous @xmath10field ( corresponding to the untwisted variational problem ) and external " torsion @xmath284 which might or might not be present . consequently , we regard @xmath1 to be part of the intrinsic data of an integrable structure . to see how the torsion encodes the geometry , we state the following proposition ( see for instance @xcite ) . [ g2tor ] for any @xmath0structure with stable form @xmath171 there exist unique differential forms @xmath382 , @xmath383 , @xmath384 , and @xmath385 so that the differentials of @xmath171 and @xmath386 are given by @xmath387 here , @xmath388 and @xmath389 denote the the bundles associated with the irreducible @xmath0modules @xmath390 and @xmath391 @xcite . to specify the torsion tensor of a connection is in general not sufficient to guarantee its uniqueness . this is true for @xmath0connections with skew symmetric torsion . using the notation of the previous proposition , we can assert the following result . [ t_comp]@xmath392 @xcite , @xcite for a @xmath0structure with stable form @xmath171 the following statements are equivalent : \(i ) the @xmath0structure is _ integrable _ , i.e. @xmath393 . \(ii ) there exists a unique linear connection @xmath377 whose torsion tensor @xmath394 is skew and which preserves the @xmath0structure , i.e. @xmath395 . the torsion can be expressed by @xmath396 moreover , the clifford action of the torsion 3form on the induced spinor @xmath397 is @xmath398 using additional subscripts @xmath399 to indicate the torsion forms of @xmath22 , equations ( [ tor ] ) and ( [ comp ] ) read @xmath400 and @xmath401 in view of theorem [ integrability ] we can use the dilatino equation to relate the torsion components to the additional parameters @xmath348 and @xmath301 . we have @xmath402 if the structure is even and @xmath403 if the structure is odd . [ torsioncomp ] if the generalised structure is weakly integrable , then there exist two linear connections @xmath22 preserving the @xmath347structure with skew torsion @xmath23 . these connections are uniquely determined . if the structure is weakly integrable and of even type , then @xmath404 and @xmath405 where @xmath406 denotes the projection of @xmath1 onto @xmath407 . moreover , the torsion can be expressed by the formula @xmath408 if the structure is weakly integrable and of odd type , then @xmath404 and @xmath409 the torsion can be expressed by the formula @xmath410 we obtain the formulae for strongly integrable case if we set @xmath328 . conversely , if we are given two @xmath0structures defined by the stable forms @xmath411 and @xmath412 inducing the same metric , a constant @xmath301 and a function @xmath21 such that ( [ tor1 ] ) or ( [ tor2 ] ) defines a closed 3form @xmath1 , then the corresponding spinors @xmath413 satisfy the integrability condition of theorem [ integrability ] and hence define a ( weakly ) integrable generalised @xmath0structure ( of even or odd type ) . the previous discussion has a striking consequence . assume @xmath216 to be compact and endowed with a weakly integrable structure of even type . then ( [ tor1 ] ) and stokes theorem imply @xmath414 here we have used that @xmath415 and that the projection of @xmath1 on @xmath416 is given by @xmath417 . the same identity holds for odd structures . since the left hand side is strictly positive unless @xmath418 , we need @xmath419 . as a result , we obtain the following no go theorem , generalising a similar statement in @xcite . [ vanishingg2 ] if @xmath216 is compact and carries an integrable generalised @xmath0structure , then @xmath320 if and only if @xmath328 . in this case the spinors @xmath413 are parallel with respect to the levi civita connection . moreover , a weakly integrable structure has necessarily non trivial torsion . in case of strong integrability , it follows in particular that the underlying topological generalised @xmath0structure can not be exotic . if the spinors were to be linearly dependent at one point , covariant constancy would imply global linear dependency and we would have an ordinary manifold of holonomy @xmath0 . if the two spinors are linearly independent at some ( and hence at any ) point , then the holonomy reduces to an @xmath163principal fibre bundle which is the intersection of the two @xmath0structures . in this case , @xmath216 is locally isometric to @xmath420 where @xmath421 is a calabi yau 3fold . next we compute the ricci tensor . to begin with , let @xmath422 and @xmath423 denote the ricci tensors associated with the levi civita connection and the connections @xmath22 . generally speaking , the ricci tensor @xmath424 associated with a @xmath204invariant spinor @xmath397 and a @xmath204preserving , metric linear connection @xmath377 with skew torsion is determined by the following relation . @xmath392 @xcite the ricci tensor associated with @xmath377 is determined by @xmath425 and relates to the metric ricci tensor through @xmath426 [ riccitensor ] the ricci tensor of an integrable generalised @xmath0structure is given by @xmath427 where @xmath428 is the hessian of the dilaton @xmath21 . it follows that the scalar curvature of the metric @xmath77 is @xmath429 where @xmath430 is the riemannian laplacian . according to the previous proposition we obtain @xmath431 and it remains to compute the ricci tensors @xmath432 and @xmath433 . since @xmath22 preserves the @xmath347structure , we derive @xmath434 from the dilatino equation , and therefore @xmath435 now consider a frame that satisfies @xmath436 at a fixed point , or equivalently , @xmath437 . as the connections @xmath22 are metric , we obtain @xmath438 the hessian evaluated in this basis is just @xmath439 , whence the result . if the dilaton is constant , then @xmath440 and the underlying @xmath347structures are co calibrated , i.e. @xmath441 . consequently , the ricci tensors @xmath423 of @xmath22 vanish . we then appeal to theorem 5.4 of @xcite which translated to our context asserts that if @xmath442 vanishes , the levi civita connection reduces to the underlying @xmath347structure . the metric @xmath77 of a strongly integrable generalised @xmath0structure is ricci flat if and only if the dilaton @xmath21 is constant . in particular , any strongly integrable generalised @xmath0structure which is homogeneous is ricci flat . the following example illustrates how restrictive the assumption @xmath415 really is . it defines a compact generalised @xmath0structure with constant dilaton and non trivial @xmath1 such that @xmath443 , @xmath444 , but @xmath445 . consider the 6dimensional nilmanifold @xmath446 associated with the lie algebra @xmath447 spanned by the orthonormal basis @xmath448 and determined by the relations @xmath449 we let @xmath450 and endow @xmath216 with the product metric @xmath451 . on @xmath446 we choose the @xmath163structure coming from @xmath452 which induces a generalised @xmath0structure on @xmath216 with @xmath453 ( cf . section [ geng2stru ] ) . we put @xmath454 . obviously , @xmath445 holds . writing @xmath455 and @xmath456 the equations @xmath443 and @xmath444 are equivalent to @xmath457 by design , @xmath458 and @xmath459 moreover , @xmath460 and @xmath461 finally , we have @xmath462 in the same vein as generalised @xmath0structures we can develop a theory of generalised @xmath16structures associated with @xmath30 in @xmath463 . we content ourselves with an outline of this theory . mutatis mutandis the proofs translate without too much difficulty from the generalised @xmath0case the only new feature to take into account is the chirality of the spinors . a more detailed exposition can be found in @xcite . as before , a generalised @xmath16structure gives rise to two unit spinors @xmath14 and @xmath15 in the associated irreducible @xmath464representations @xmath465 and @xmath466 ( the subscript does not indicate the chirality of the spinors ) . tensoring those induces a @xmath30invariant spinor which is given by either an even or an odd form according to the chirality of the spinors @xmath14 and @xmath15 . this leads to the notion of generalised @xmath16structures of _ even _ or _ odd _ type for spinors of equal or opposite chirality . the box operator is now a map @xmath467 for which the ( anti)self duality property @xmath190_b=(-1)^{ev , od}\box_{g , b}[\psi_+\otimes\psi_-]_b\ ] ] holds . to any @xmath30invariant spinor we can then associate the volume form @xmath468 and define the dilaton @xmath21 by @xmath469 . note that such a spinor is not stable . a _ topological generalised @xmath16structure _ over an 8manifold @xmath216 is a topological @xmath30reduction of the @xmath463principal fibre bundle associated with @xmath32 . it is said to be of _ even _ or _ odd type _ according to the parity of the @xmath30invariant spinor @xmath19 . we will denote this structure by the pair @xmath217 and call @xmath19 the _ structure form_. equivalently , such a structure can be characterised by the following data : * an orientation * a metric @xmath77 * a 2form @xmath9 * a scalar function @xmath21 * two unit half spinors @xmath220 of either equal ( even type ) or opposite chirality ( odd type ) such that @xmath221_b=\rho.\ ] ] the normal form description is computed as in the @xmath0case . there exists an orthonormal basis @xmath470 such that @xmath162 & = & \phantom{+}\cos(a)+\sin(a)(e_{12}-e_{34}-e_{56}+e_{78})+\\ & & + \cos(a)(e_{1234}+e_{1246}-e_{1278}+e_{1357}+e_{1368}+e_{1458}-e_{1467}-\\ & & -e_{2358}+e_{2378}+e_{2457}+e_{2468}-e_{3467}+e_{4578}+e_{5678})+\\ & & + \sin(a)(e_{1358}-e_{1367}-e_{1457}-e_{1468}+e_{2357}+e_{2368}+e_{2458}-e_{2467})+\\ & & + \sin(a)(-e_{123456}+e_{123478}+e_{125678}-e_{345678})+\cos(a)e_{12345678},\end{aligned}\ ] ] where @xmath471 and the spinors are of equal chirality ( even case ) . in the odd case , we have @xmath162 & = & -e_1 + e_{234}+e_{256}-e_{278}+e_{357}+e_{368}+e_{458}-e_{467}+\\ & & { } + e_{12358}-e_{12367}-e_{12457}-e_{12468}+e_{13456}-e_{13478}-e_{15678}+e_{2345678}.\end{aligned}\ ] ] if the spinors are not parallel , we can express the homogeneous components of @xmath17 $ ] in terms of the invariant forms of the intersection @xmath472 ( even case ) or @xmath473 ( odd case ) . we find @xmath168 = c+s\omega+c(\psi_+-\frac{1}{2}\omega^2)-s\psi_-\frac{s}{6}\omega^3+cvol_g\ ] ] where @xmath166 is the khler form and @xmath167 the real and imaginary parts of the holomorphic volume form stabilised by @xmath474 . in the odd case , we have @xmath168 = -\alpha+\varphi-\alpha\wedge\psi+\frac{1}{7}\varphi\wedge\psi\ ] ] where @xmath165 denotes the dual of the unit vector in @xmath1 which is stabilised by @xmath0 , @xmath171 the invariant stable 3-form on the complement and @xmath475 . existence of generalised @xmath16structures follows from the existence of either a unit spinor ( even structures ) or a unit spinor and a unit vector field ( odd structure ) , both of which is classical @xcite . an 8manifold @xmath216 carries an even topological generalised @xmath16structure if and only if @xmath216 is spin and @xmath476 . \(ii ) a differentiable 8manifold @xmath216 carries an odd topological generalised @xmath16structure if and only if @xmath216 is spin , has vanishing euler class and satisfies @xmath477 . for a generalised @xmath16structure of even type we can discuss classification issues as in section [ topgeng2 ] . the sphere bundle associated with @xmath465 has fibre isomorphic to @xmath234 and an 8dimensional base , so that two transverse sections will intersect in a curve . we meet the first obstruction for the existence of a vertical homotopy in @xmath251 , which by poincar duality trivially vanishes if @xmath478 ( e.g. if @xmath216 is simply connected ) . since @xmath479 , the second obstruction lies in the top cohomology module @xmath480 the stable homotopy group @xmath481 is isomorphic to the framed cobordism group of @xmath482manifolds . it is conceivable that the @xmath104class is the framed ( or spin ) cobordism class of the 1manifold where the two sections coincide . the tangent bundle of the 8sphere is stably trivial and therefore all the pontrjagin classes vanish . since the euler class is non trivial , there exists no generalised @xmath16structure on @xmath483 . however , they do exist on manifolds of the form @xmath484 for @xmath485 spinnable . for instance , take @xmath485 to be the 7sphere @xmath234 . then the tangent bundle of @xmath216 is trivial and so is the sphere bundle @xmath249 associated with the spinor bundle . hence @xmath486 $ ] which contains the set @xmath487=\pi_7(s^7)={\mathbb{z}}$ ] . choosing a non trivial homotopy class in @xmath488 which we extend trivially to @xmath489 defines an exotic generalised @xmath16structure . since the structure form is not stable , we can not setup a variational problem . we therefore follow the analogy of the classical case and impose ad hoc the strong integrability condition . a topological generalised @xmath16structure @xmath217 is said to be _ integrable with respect to a closed 3form _ @xmath284 if and only if @xmath490 \(i ) consider an 8manifold @xmath216 endowed with a @xmath16invariant 4form @xmath491 and associated riemannian volume form @xmath492 . the structure form of the induced straight structure ( necessarily of even type ) is given by the @xmath10field transform of @xmath493 which follows from the normal form description above . we want to solve @xmath494 which is equivalent to @xmath495 . consequently , the holonomy of @xmath496 reduces to @xmath16 . \(ii ) examples of odd type are provided by product manifolds of the form @xmath497 , where @xmath485 carries a @xmath0structure induced by the stable 3form @xmath171 . let @xmath498 be a closed 4form on @xmath216 . the spinor @xmath499 defines a generalised @xmath16structure of odd type which is @xmath375closed if and only if @xmath319 , @xmath320 and @xmath321 , @xmath500 , i.e. the @xmath0structure on @xmath446 is calibrated . \(iii ) manifolds with holonomy contained in @xmath16 can be easily built out of a trivial @xmath27bundle over a 7manifold with holonomy @xmath0 . this easily generalises to our context where a strongly integrable generalised @xmath0structure @xmath501 induces an integrable generalised @xmath16structure of even type @xmath502 . [ spin7integrability ] a generalised @xmath16structure @xmath217 is integrable if and only if @xmath343_b=\rho$ ] satisfies ( with @xmath317 ) @xmath503 and @xmath504 using theorem [ spin7integrability ] we can discuss the torsion of the underlying @xmath16structures defined by the invariant 4forms @xmath505 in the same way as in the @xmath0case . using results from @xcite , we obtain : if the generalised @xmath16structure is integrable , then @xmath506 and @xmath507 where @xmath508 denotes the projection of @xmath1 onto @xmath509 , the bundle associated with the irreducible @xmath510representation space @xmath511 in @xmath512 @xcite . conversely , if we are given two @xmath510invariant forms inducing the same metric @xmath77 , a function @xmath21 and a closed 3form @xmath1 such that ( [ torspin7cor ] ) and ( [ domegaspin7 ] ) hold , then the corresponding spinors @xmath413 satisfy theorem [ spin7integrability ] and hence define an integrable generalised @xmath16structure . as in the @xmath0case , we then deduce the following results . [ vanishingspin7 ] if @xmath216 is compact and carries an integrable generalised @xmath16structure , then @xmath320 . consequently , the spinors @xmath413 are parallel with respect to the levi civita connection . the ricci tensor and the scalar curvature of an integrable generalised @xmath16structure are given by @xmath513 and @xmath514 an integrable generalised @xmath16structure is ricci flat if and only if the dilaton @xmath21 is constant . in particular , any homogeneous integrable generalised @xmath16structure is ricci flat . again the condition @xmath415 is crucial . in conjunction with @xmath515 of the previous example , the compact @xmath0structure discussed at the end of section [ ricci ] gives trivially rise to an instance of a compact generalised @xmath16structure of even type that satisfies @xmath443 , but @xmath445 . for an odd example , just take a compact calibrated @xmath0manifold which , for instance , can be built out of the nilmanifold @xmath446 considered above . it follows that @xmath516 ( proposition [ g2tor ] ) , so that @xmath443 for @xmath517 and @xmath19 defined as in @xmath518 of the previous example . the torsion 3form of a compact calibrated @xmath0manifold can never be closed . type iia and iib string theory are interrelated by the so called @xmath1duality . formally speaking , @xmath1duality transforms the data @xmath519 consisting of a generalised metric @xmath94 and the dilaton @xmath21 , all living on a principal @xmath27bundle @xmath520 with connection form @xmath521 , into a generalised metric @xmath522 and a dilaton @xmath523 over a _ new _ principal @xmath27fibre bundle @xmath524 with connection form @xmath525 . in the physics literature , the coordinate description of this dualising procedure is known under the name of _ buscher rules_. a neat mathematical formulation particularly apt for applications in the setting of generalised geometry was given in @xcite . let @xmath526 denote the dual vertical vector field of @xmath521 , i.e. @xmath527 . consider its curvature 2form @xmath528 which we regard as a 2form on @xmath216 so @xmath529 . we assume to be given a closed , integral and @xmath27invariant 3form @xmath1 such that the 2form @xmath530 defined by @xmath531 is also integral . in practice , we will assume @xmath320 so that this condition is automatically fulfilled . integrality of @xmath530 ensures the existence of another principal @xmath27bundle @xmath532 , the _ @xmath1dual _ of @xmath533 defined by the choice of a connection form @xmath525 with @xmath534 . writing @xmath535 for a 3form @xmath536 , we define the @xmath1dual of @xmath1 by @xmath537 here and from now on , we ease notation and drop the pull back @xmath538 . to make contact with our situation , consider an @xmath27invariant structure form @xmath19 which we decompose into @xmath539 the _ @xmath1dual of _ @xmath19 is defined to be @xmath540 in particular , @xmath1duality reverses the parity of forms and maps even to odd and odd to even forms . it is enacted by multiplication with the element @xmath541 on @xmath19 followed by the substitution @xmath542 . the crucial feature of @xmath1duality is that it preserves the @xmath50orbit structure on @xmath543 . to see this , we decompose @xmath544 where @xmath545 denotes the vertical summand @xmath546 of @xmath547 which is spanned by @xmath526 , and similarly for @xmath521 and their @xmath1duals . then consider the map @xmath548 defined with respect to this splitting by @xmath549 it satisfies @xmath550 for any @xmath551 and in particular , @xmath552 . hence this map extends to an isomorphism @xmath553 , and any orbit of the form @xmath554 gets mapped to an equivalent orbit @xmath555 where @xmath204 and @xmath556 are isomorphic as abstract groups . as an illustration of this , consider a generalised @xmath0structure over @xmath533 with structure form @xmath557 and companion @xmath558 . these have @xmath1duals @xmath559 and @xmath560 . since @xmath19 and @xmath307 have the same stabiliser @xmath204 inside @xmath561 , it follows that @xmath559 and @xmath560 are stabilised by the same @xmath556 inside @xmath562 , which is isomorphic to @xmath7 . by invariance , @xmath560 and @xmath563 coincide up to a constant which we henceforth ignore . the integrability condition transforms as follows : we put this machinery into action as follows . start with a non trivial principal @xmath27fibre bundle @xmath565 which admits a metric of holonomy @xmath0 or @xmath16 and let @xmath320 . the resulting straight structure is strongly integrable and so is its dual , but according to the @xmath1duality rules , we acquire non trivial torsion given by @xmath566 . local examples of such @xmath0structures exist in abundance @xcite . in conjunction with @xmath515 of the first example in section [ spin7theory ] , this gives an @xmath27invariant generalised @xmath16structure of even type with integral , @xmath27invariant torsion @xmath1 . contracting with @xmath567 yields @xmath568 , hence the @xmath1dual defines an integrable generalised @xmath16structure of odd type with @xmath569 . as a further application of this formalism , we note that a principal @xmath27fibre bundle @xmath570 whose holonomy reduces to an @xmath27invariant @xmath0 or @xmath16structure must be trivial if the base @xmath216 is compact . indeed , if @xmath521 is a connection form on @xmath533 and @xmath320 , then the @xmath1dual defines a strongly integrable generalised structure with torsion @xmath566 . but this vanishes as a consequence of corollaries [ vanishingg2 ] and [ vanishingspin7 ] , hence @xmath571 which implies the triviality of @xmath533 . . stable forms and special metrics . in m. fernndez and j. a. wolf ( ed . ) . _ global differential geometry : the mathematical legacy of alfred gray ( bilbao , 2000 ) _ , contemp . math . * 288 * , 7089 , 2001 , math.dg/0107101 .	we define new riemannian structures on 7manifolds by a differential form of mixed degree which is the critical point of a ( possibly constrained ) variational problem over a fixed cohomology class . the unconstrained critical points generalise the notion of a manifold of holonomy @xmath0 , while the constrained ones give rise to a new geometry without a classical counterpart . we characterise these structures by means of spinors and show the integrability conditions to be equivalent to the supersymmetry equations on spinors in type ii supergravity theory with bosonic background fields . in particular , this geometry can be described by two linear metric connections with skew symmetric torsion . finally , we construct explicit examples by introducing the device of @xmath1duality .
the discovery of iron - based superconductors is the most significant breakthrough in the condensed matter physics in recent years@xcite . the iron - pnictides / chacogenides are close to magnetism and it is widely believed that the spin - fluctuation is responsible for the pairing of the superconducting electrons@xcite . compared with the iron - based compounds , the nickel - based systems have much lower critical temperature , for example , laonip ( @xmath4@xcite , laonias ( @xmath5@xcite , bani@xmath0p@xmath0 ( @xmath6@xcite , bani@xmath0as@xmath0 ( @xmath7@xcite , usually lower than 5 k even upon doping@xcite . to date , no evidence for ordered or even fluctuated magnetism associated with ni in proximity to superconductivity is found . moreover , the majority of evidence suggests that the ni - based systems can be understood in context of conventional electron - phonon theory@xcite . recently , there nickel chalcogenides kni@xmath0se@xmath0 ( @xmath8@xcite , kni@xmath0s@xmath0 ( @xmath9@xcite , and tlni@xmath0se@xmath0(@xmath10@xcite were reported , in which superconductivity appears to involve heavy - fermion behavior below a coherent temperature @xmath11@xmath1220 k. in contrast to k@xmath13fe@xmath14se@xmath0@xcite , tlni@xmath0se@xmath0compound is homogeneous without ni vacancy or phase separation@xcite . considering the monovalent tl or k in the stoichiometry composition , the effective valance of ni in such compound is `` 1.5 + '' . the mixed valency of @xmath15 was proposed to induce the heavy effective band mass@xcite . however , very recently , angle - resolved photoemission spectroscopy ( arpes ) measurement@xcite reveals that tlni@xmath0se@xmath0shares a universal band structure with bafe@xmath0as@xmath0@xcite and baco@xmath0as@xmath0@xcite , with a chemical potential shifted due to more @xmath16 electrons in tlni@xmath0se@xmath0(3@xmath17 ) than in bafe@xmath0as@xmath0 ( @xmath18 and baco@xmath0as@xmath0 ( @xmath19 . therefore , it can be viewed as heavily electron doped tlfe@xmath0se@xmath0 ( @xmath20 , with reduced electronic correlations than in its cousin k@xmath13fe@xmath14se@xmath0@xcite . furthermore , the camelback - shaped band at z point gives rise to a pronounced van hove singularity ( vhs ) near fermi energy ( @xmath21 ) , which provides a natural explanation to the heavy electrons behavior inferred from the electronic heat and the upper critical field measurements@xcite in this weakly correlated system . more specially , the neutron pair - distribution - function ( pdf ) analysis in a combined high - resolution synchrotron x - ray diffraction and neutron scattering study in kni@xmath0se@xmath0 reveals that the local charge density wave ( cdw ) state is present up to 300 k but disappears on cooling in the `` heavy - fermion '' state below @xmath11@xcite . the effect is opposite to what is typically observed . moreover the arpes did not observe any anomaly in the spectra upon entering the fluctuating cdw state@xcite . to understand the unique behavior , it is highly desirable to investigate the charge dynamics in the whole temperature range below 300 k. optical spectroscopy is a powerful bulk sensitive technique with high energy resolution , which is widely used to probe the carrier dynamics and possible gap formation of an electronic system . in this work , we present a detailed optical spectroscopy study on tlni@xmath0se@xmath0single crystals . our optical data reveals that tlni@xmath0se@xmath0is a good metal with a distinct plasma reflectance edge , similar to its isostructure bani@xmath0as@xmath0 . the overall frequency dependent reflectivity @xmath2 is gradually suppressed near 2500 @xmath22with decreasing temperature , and a pseudogap feature appears at any temperature below 300 k. the weak gap structure is attributed to the local cdw order or fluctuations . our measurement indicates that the local cdw order / fluctuations are further enhanced at the lowest temperature , which is different from the neutron pdf analysis on kni@xmath0se@xmath0 in its `` heavy fermion state '' . in combination with the earlier arpes results , we suggest that the fluctuating cdw instability is driven by the presence of saddle points in band structure which leads to the van hove singularity close to the fermi level . single crystals of tlni@xmath0se@xmath0were grown by the self - flux method@xcite . their structure , transport , magnetic and thermodynamic properties were already reported@xcite . the frequency - dependent reflectance spectra at different temperature were measured by bruker ifs 113v and 80v spectrometers in the frequency range from 30 to 20 000 @xmath22(4 [email protected] ev ) , and then the reflectance was extended to 50 000 @xmath22(@xmath126.1 ev ) at room temperature with a grating - type spectrometer . an _ in situ _ gold and aluminum overcoating technique was used to obtain the reflectivity r(@xmath23 ) . considering the small size of our samples , the data below 100 @xmath22are cut off for reliability . the real part of conductivity @xmath3 is obtained by the kramers - kronig transformation of r(@xmath23 ) . the hagen - rubens relation was used for low frequency extrapolation ; at high frequency side , an extrapolation method using x - ray atomic scattering functions was applied to generate the high - frequency reflectivity@xcite . figure 1 shows the reflectance spectra of tlni@xmath0se@xmath0single crystals over a broad energy scale at various temperatures . the value of r(@xmath23 ) approaches unity towards zero energy and shows an increase with decreasing temperature , indicating a good metallic behavior . by lowering the temperature , we do not see any sharp changes in the optical spectra but rather weak suppression of r(@xmath23 ) near 2500 @xmath22 , signaling the opening of a partial gap ( pseudogap ) . with increasing @xmath23 , r(@xmath23 ) drops quickly near 7000 @xmath22 , known as the screened plasma edge . the relativity high edge position reveals a high carrier density , which is similar to isostructural compound , bani@xmath0as@xmath0@xcite . there is another edge structure at about 15 000 @xmath22which is caused by an interband transition . the reflectivity becomes roughly temperature independent at higher frequencies . the inset of fig . 1 shows r(@xmath23 ) data up to 50 000 @xmath22 in a logarithmic scale at 300 k. ) _ in the frequency range from 100 to 20 000 @xmath22 . the inset shows r(@xmath23 ) data up to 50 000 @xmath22on a logarithmic scale at 300 k.,width=307 ] figure 2 shows the real part of conductivity spectra below 12 000 @xmath22at different temperatures . the drude - type conductivity is observed for all spectra at low frequency and several temperature - independent interband transitions could be also well resolved , as shown in the inset of figure 2 . the optical conductivity between 500 and 10 000 @xmath22is gradually suppressed and a broad peak - like feature becomes more and more obvious at 2800 @xmath22as the temperature decreases . the spectral line shape is similar to that of bafe@xmath0as@xmath0@xcite across the sdw transition though the feature is much weaker . it seems to suggest that there should also be a density - wave gap opening in tlni@xmath0se@xmath0 . furthermore , this pseudogap feature is most prominent at low temperature and can be resolved even in the 300 k data . to quantify the temperature evolution of optical conductivity , especially the low frequency components , we use a simple drude - lorentz model to decompose optical conductivity spectra@xcite : @xmath24 where @xmath25 and @xmath26 are the plasma frequency and the relaxation rate of each conduction band , while @xmath27 , @xmath28 and _ s@xmath29 _ are the resonance frequency , the damping and the mode strength of each lorentz oscillator , respectively . this model includes drude and lorentz terms , which approximately capture the contribution by free carriers and interband transitions . both transport@xcite and arpes@xcite results suggest the multiband character of tlni@xmath0se@xmath0 , therefore we applied two drude components analysis here , similar to the other members of `` 122 '' system@xcite . even though the drude - lorentz fit is somewhat arbitrary and the accuracy of the derived fit parameters is uncertain , it will not modify the key outcomes of the analysis . at different temperatures . the inset shows _ @xmath3 _ over broad frequencies up to 50 000 @xmath22on a logarithm scale.,width=307 ] figure 3 illustrates the conductivity spectra at 300 k and 10 k together with the drude - lorentz fitting components at 10 k for tlni@xmath0se@xmath0 . in order to reproduce the optical conductivity below 50 000 @xmath22at 300 k , two drude components , a narrow and a broad one , and four lorentz components have to be used . furthermore , an additional midinfrared peak is necessary near about 2800 @xmath22at any temperature below 300 k. we are mainly concerned about the evolution of the low energy conductivity , therefore the fitting parameters for the the two drude components are shown in the table 1 for different temperatures . we also show the error bars of the fitting parameters corresponding to an absolute error of 0.5% in reflectivity . we find that the two drude components narrow with decreasing temperature due to the metallic response . the plasma frequency @xmath30 keeps roughly unchanged , while @xmath31 decreases suggesting that the gapping of the fermi surfaces mainly appears in the broad drude component . the lorentz 5 oscillator strength @xmath32 increases continuously with decreasing temperature , indicating the progressive formation of the absorption peaks just above the pseudogap . we use the formula @xmath33 to estimate the overall plasma frequency , then we get @xmath34 36360 @xmath22at 300 k , and 32400 @xmath22at 10 k , respectively . therefore , the ratio of the square of the plasma frequency at 10 k to that of 300 k is about 0.80 . it is well known that @xmath35 , where _ n _ is the carrier density and @xmath36 is the effective mass . if we assume that the effective mass of itinerant carriers would not change with temperature , the optical data reveals that roughly 20% of the itinerant carriers is lost due to the pseudogap opening at the fermi surface . at 300 and 10 k with the drude - lorentz fits shown at the bottom.,width=307 ] _ in tlni@xmath0se@xmath0and @xmath37 . the inset shows the model ratio of the low temperature integrated sw over that of the model optical conductivity at 10 k.,width=307 ] * 6c22mmc + t(k)&@xmath30&@xmath38&@xmath31&@xmath39&@xmath32&@xmath40 + + 300 & 18500 & 390 & 31300 & 2600 & 0 & 36360 + 200 & 18500 & 310 & 29400 & 2200 & 9500 & 34740 + 100 & 18500 & 270 & 27700 & 1900 & 12000 & 33310 + 10 & 18500 & 200 & 26600 & 1500 & 14100 & 32400 + model & 18500 & 217 & 30100 & 1500 & 0 & 35300 + error bars & @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath41 & - + to further characterize the spectral evolution , we plot the integrated spectral weight of tlni@xmath0se@xmath0at 10 k , as shown in figure 4 , where the low - temperature integrated spectral weight has been normalized to the integrated spectral weight at room temperature . the optical spectral weight is defined as @xmath45 . for the convenience of comparison , we also present the low - temperature integrated spectral weight of bafe@xmath0as@xmath0@xcite . for bafe@xmath0as@xmath0 , the strong dip below 1000 @xmath22represents the formation of sdw gap in the magnetic ordered state@xcite , while the second dip near 3000 @xmath22reflects the unconventional sw transfer at high energy which seems to be a common feature in iron - pnictides / chacogenides@xcite . nevertheless , for tlni@xmath0se@xmath0 , the temperature - dependent spectral change are very different . the sw ratio exceeds 1 before it recovers to unity suggesting that the sw is transferred from high to low energy at low temperature . we should remark that most of the sw below 6000 @xmath22(the minimum of optical conductivity ) would come from the intraband transitions . as the temperature decreases , the sw transformation from high to low energy is induced by the drude components narrowing . this is the typical optical response of a metallic material . on the other hand , the pseudogap develops below 2000 @xmath22at low temperature . however , we note that the effects of the pseudogap formation to the sw transfer are minimal . it is the metallic response that leads to the sw transfer to low energy . in other words , the pseudogap is too weak to change the sw evolution in such a good metal . notably , the sw of the mid - infrared peak could not fully compensate for the loss of the low - energy drude component , and the recovery of sw extends to fairly high energy scale . furthermore , the high energy sw transfer above 8000 @xmath22is almost undistinguishable in tlni@xmath0se@xmath0 . this may be related to the weaker correlation or hund s coupling effect in tlni@xmath0se@xmath0as compared to @xmath37 , in agreement with the arpes results@xcite . in order to make the pseudogap - like suppression more clear , we also plot a model ratio of the low - temperature integrated sw to the model integrated spectral weight at 10 k. assuming that tlni@xmath0se@xmath0is a pure metal without the pseudogap feature , the model optical conductivity at 10 k could be obtained according to the drude - lorentz model with two drude components and four lorentz components , as the situation at 300 k . the parameters of the model optical conductivity are listed in table 1 , where @xmath38 is set to 217 @xmath22 in order to keep the same dc optical conductivity as the real data and @xmath31 is set to 30100 @xmath22to conserve the total spectral weight at 10 k. we then normalize the integrated sw at low temperature sw(10 k ) to the model value msw(10 k ) . as shown in the inset of fig . 4 , the ratio decrease below @xmath461700 @xmath22 , suggesting that the sw is depleted due to opening of the pseudogap , and the suppressed sw is transferred to higher energy , leading to the gradually recovery of the sw above the pseudogap . this is a typical behavior of gap opening induced sw transformation . now we can summarize the main finding of our optical data : the overall optical spectra are similar to those observed in its isostructure @xmath1 . corresponding to the weak suppression of @xmath2 , the real part of the conductivity @xmath3 shows a suppression roughly below 2000 @xmath22 , which leads to a peak above this energy . the mid - infrared peak is present at all measurement temperatures below 300 k , and becomes more and more obvious with decreasing temperature , indicating the gradual formation of pseudogap . at the same time , irrespective of the uncertainly , the overall plasma frequency decreases , implying the partial gap opening on the fermi surface , which gives a further proof of the pseudogap formation . in particular , the sw transformation to higher energy induced by pseudogap could be clearly observed in the model ratio , where we use the metallic response at 10 k as a reference . as we have mentioned above , for kni@xmath0se@xmath0 or kni@xmath0s@xmath0 , the neutron pdf analysis suggested that the local cdw fluctuations are present at high temperature but disappear upon further cooling@xcite . however , our optical data of tlni@xmath0se@xmath0single crystals reveals the progressive formation of pseudogap in the midinfrared range , especially below @xmath11 . to understand the origin of pseudogap in tlni@xmath0se@xmath0crystals , there are also a few other experimental facts which should be taken into account . first , to date no evidence for magnetic transition was found on such nickel chalcogenides . second , no static density wave , such as cdw or sdw , has been observed . third , the anisotropy in tlni@xmath0se@xmath0is rather small , although the compound has a layer structure . it looks like that the pseudogap in tlni@xmath0se@xmath0might be originated from local cdw fluctuations , in agreement with earlier neutron measurement on kni@xmath0se@xmath0 and kni@xmath0s@xmath0@xcite . moreover , the local cdw fluctuation was suggested to be entirely dynamic and/or spatially incoherent@xcite , in contrast to the coherence cdws observed in structurally related compounds such as _ 2h_-typed transition metal dichalcogenides@xcite . with such cdw fluctuation , there could be a partial gap - opening in k - space on the fermi surface , which could be related to the pseudogap feature in tlni@xmath0se@xmath0 . however , different from kni@xmath0se@xmath0 and kni@xmath0s@xmath0 where the fluctuated cdw order tends to disappear upon entering the heavy electron state at very low temperature , the optical measurement on tlni@xmath0se@xmath0indicates that the partial gap feature is further enhanced at the lowest temperature . an earlier angle - resolved photoemission spectroscopy measurement on kni@xmath0se@xmath0 crystals suggested the quasi - two - dimensional nature of the fermi surface and that the partial nested fermi surface might be responsible for the local cdw fluctuations@xcite . however , a more recent arpes measurements on tlni@xmath0se@xmath0revealed more complicated and three dimensional fermi surface shape . thus , a naive picture for a fs nesting scenario is not applicable here on tlni@xmath0se@xmath0 . furthermore , a camelback - shaped band near fermi energy at _ z _ point was identified , in particular , the existence of four flat part of the dispersion very close to @xmath21 would give rise to a pronounced van hove singularity ( vhs)@xcite . on this basis , the cdw instability can be naturally understood in the frame of `` saddle - point '' cdw mechanism , proposed by rice and scott@xcite . the four saddle points near _ z _ point in tlni@xmath0se@xmath0gives a divergent contribution to lindhard response function @xmath47 , where @xmath48 is a wave vector connecting two saddle points , leading to a cdw instability . we have performed optical spectroscopy study on tlni@xmath0se@xmath0over a broad frequency range at various temperatures . both the suppression in @xmath2 and the peaklike feature in @xmath3 suggest the progressive formation of a pseudogap feature in the midinfrared range with decreasing temperatures , which might be originated from the dynamic local fluctuation of charge - density - wave ( cdw ) instability . the cdw instability in tlni@xmath0se@xmath0may not be driven by fermi surface nesting , but by the saddle points mechanism , due to the exitance of van hove singularity very close to the fermi energy . b. zhou , m. xu , y. zhang , g. xu , c. he , l. x. yang , fei chen , b. p. xie , x. y. cui , masashi arita , kenya shimada , hirofumi namatame , masaki taniguchi , x. dai , and d. l. feng , phys . b * 83 * , 035110 ( 2011 ) . s. ideta , t. yoshida , m. nakajima , w. malaeb , h. kito , h. eisaki , a. iyo , y. tomioka , t. ito , k. kihou , c. h. lee , y. kotani , k. ono , s. k. mo , z. hussain , z .- x . shen , h. harima , s. uchida , and a. fujimori , phys . b * 89 * , 195138 ( 2014 ) . n. xu , c. e. matt , p. richard , a. van roekeghem , s. biermann , x. shi , s .- f . wu , h. w. liu , d. chen , t. qian , n. c. plumb , m. radovic , hangdong wang , qianhui mao , jianhua du , minghu fang , j. mesot , h. ding , m. shi , phys . b * 92 * , 081116 ( 2015 ) .	quasi - two - dimensional nickel chalcogenides tlni@xmath0se@xmath0is a newly discovered superconductor . we have performed optical spectroscopy study on tlni@xmath0se@xmath0single crystals over a broad frequency range at various temperatures . the overall optical reflectance spectra are similar to those observed in its isostructure @xmath1 . both the suppression in @xmath2 and the peaklike feature in @xmath3 suggest the progressive formation of a pseudogap feature in the midinfrared range with decreasing temperatures , which might be originated from the dynamic local fluctuation of charge - density - wave ( cdw ) instability . we propose that the cdw instability in tlni@xmath0se@xmath0is driven by the saddle points mechanism , due to the existence of van hove singularity very close to the fermi energy .
the release of the third year data from the wilkinson microwave anisotropy probe ( _ wmap _ ) has further improved the precision with which we can constrain the cosmological parameters from the shape of the cosmic microwave background ( cmb ) anisotropies @xmath5 @xcite . the _ planck _ satellite , scheduled for launch in 2008 @xcite , will provide even higher precision @xmath5 values and data down to smaller angular scales ( @xmath7 ) . higher precision in the observations requires increased accurarcy from the theoretical calculations , in order for the correct cosmological parameters to be extracted . it now seems crucial to obtain the @xmath5s down to at least the 1 percent level over a wide range of @xmath8 . cmbfast @xcite is the most commonly used boltzmann code for calculating the @xmath5s , and it gives consistent results with other independent codes ( see * ? ? ? * and references therein ) . the dominant uncertainty in obtaining accurate @xmath5s comes from details in the physics of recombination , for example , the ` fudge factor ' in the recfast routine @xcite . calculations of cosmological recombination were first published by @xcite and @xcite . @xcite presented the most detailed multi - level calculation and introduced a fudge factor to reproduce the results within an effective three - level model . although the multi - level calculation already gives reasonable accuracy , the required level of accuracy continues to increase , so that today any effect which is @xmath91 per cent over a range of multipoles is potentially significant . several modifications have been recently suggested to give per cent level changes in the ionization fraction and/or the @xmath5s ( see section 4 for details ) . most of these modifications have been calculated only with an effective three - level code , and so the results may be different in the multi - level calculation , since there is no thermal equilibrium assumed between the upper states . here we want to focus on one of these modifications , namely the extra forbidden transitions proposed by @xcite , which we study using a multi - level code . in the standard calculations of recombination , one considers all the resonant transitions , but only one forbidden transition , which is the 2s1s two - photon transition , and this can be included for both h and he . @xcite suggested that one should also include the two - photon transitions from higher excited s and d states to the ground state for h and hei , and also the spin - forbidden transition between the triplet @xmath10p@xmath11 and singlet ground state @xmath12s@xmath13 for hei . they showed that the recombination of both h and hei sped up in the three - level atom model . the suggested level of change is large enough to bias the determination of the cosmological parameters @xcite . the purpose of this paper is to investigate the effect of the extra forbidden transitions suggested by @xcite in the multi - level atom model without assuming thermal equilibrium among the higher excited states . the outline of the paper is as follows . in section 2 we will describe details of the rate equations in our numerical model . in section 3 we will present results on the ionization fraction @xmath4 and the anisotropies @xmath5 , and assess the importance of the addition of the forbidden transitions . other possible improvements of the recombination calculation will be discussed in section 4 . and finally in section 5 we will present our conclusions . in this paper we follow the formalism of the multi - level calculation performed by @xcite . we consider 100 levels for hi , 103 levels for hei , 10 levels for heii , 1 level for heiii , 1 level for the electrons and 1 level for the protons . for h , we only consider discrete @xmath0 levels and assume that the angular sub - levels ( @xmath8-states ) are in statistical equilibrium within a given shell . for both hei and heii , we consider all the @xmath8-states separately . the multilevel hei atom includes all states with @xmath2 and @xmath3 . here we give a summary of the rate equations for the number density of each energy level @xmath14 , and the equation for the change of matter temperature @xmath15 . the rate equation for each state with respect to redshift @xmath16 is @xmath17 + 3 n_i , \end{aligned}\ ] ] + where @xmath18 is the number density of the @xmath14th excited atomic state , @xmath19 is the number density of electrons , and @xmath20 is the number density of continuum particles such as a proton , heii , or heiii ion . additionally @xmath21 is the photo - recombination rate , @xmath22 is the photo - ionization rate , @xmath23 is the net bound - bound rate for each line transition , and @xmath24 is the hubble parameter . we do not include the collisional rates , as they have been shown to be negligible @xcite . for hei , we update the atomic data for the energy levels @xcite , the oscillator strength for resonant transitions ( * ? ? ? * in preparation ) and the photo - ionization cross - section spectrum . we use the photo - ionization cross - section given by @xcite for @xmath0@xmath2510 and @xmath8@xmath254 , and adopt the hydrogenic approximation for states with @xmath8@xmath265 @xcite . it is hard to find published accurate and complete data for the photo - ionization cross - section of hei with large @xmath0 and @xmath8 . for example , a recent paper by @xcite claimed that they had calculated the photo - ionization cross - section up to @xmath0@xmath2727 and @xmath8@xmath2726 , although , no numerical values were provided . the rate of change of matter temperature with respect to redshift is @xmath28 where @xmath29 is the radiation temperature , @xmath30 is the total number density for helium , @xmath31 is the electron mass , @xmath32 is the speed of light , @xmath33 , @xmath34 is the radiation constant and @xmath35 is the thompson scattering cross - section . @xcite considered all the resonant transitions and only one forbidden transition , namely the 2s1s two - photon transition , in the calculation of each atom , ( for hei , 2s@xmath36s@xmath37 and 1s@xmath38s@xmath37 ) . the 2s1s two - photon transition rate is given by @xmath39 where @xmath40 is the spontaneous rate of the corresponding two - photon transition , @xmath41 is the frequency between levels 2s and 1s , @xmath42 is the degeneracy of the energy level @xmath14 , @xmath43 is planck s constant and @xmath44 is boltzmann s constant . here we include the following extra forbidden transitions , which were first suggested by @xcite . the first ones are the two - photon transitions from @xmath0s and @xmath0d to 1s for h , plus @xmath45s@xmath37 and @xmath45d@xmath46 to @xmath47s@xmath37 for hei . for example , for h , we can group together the @xmath0s and @xmath0d states coming from the same level , so that we can write the two - photon transition rate as @xmath48 + here @xmath0 ( without a subscript ) is the principle quantum number of the state , @xmath49 is the total number density of the excited atoms in either the @xmath0s or @xmath0d states , and @xmath50 is the effective spontaneous rate of the two - photon transition from @xmath51 to 1s , which is approximated by the following formula @xcite : @xmath52 where @xmath53 is equal to 8.2290s@xmath54 @xcite . the latest value of @xmath53 is equal to 8.2206s@xmath54 @xcite and does not bring any noticeable change to the result . here @xmath55 is equal to 1 for @xmath0@xmath272 , and 6 for @xmath0@xmath563 . this spontaneous rate is estimated by considering only the non - resonant two - photon transitions through one intermediate state @xmath0p . @xcite ignored the resonant two - photon transition contributions , since the escape probability of these emitted photons is very low . the above formula for @xmath50 is valid up to @xmath0@xmath57@xmath58 , due to the dipole approximation used , although it is not trivial to check how good this approximate rate is . besides the 2s1s two - photon rate , only the non - resonant two - photon rates from 3s to 1s and 3d to 1s are calculated accurately and available in the literature . @xcite evaluated @xmath59 and @xmath60 by including the non - resonant transitions through the higher - lying intermediate @xmath0p states ( @xmath0@xmath56@xmath61 ) , which are equal to 8.2197s@xmath54 and 0.13171s@xmath54 , respectively . these values were confirmed by @xcite and agreed to three significant figures . using these values , we find that @xmath50 is equal to 1.484s@xmath54 , which is an order of magnitude smaller than the value from the approximated rate coming from equation ( 5 ) . the approximation given by @xcite therefore seems to be an overestimate . this leads us instead to consider a scaled rate @xmath62 , which is equal to @xmath50 multiplied by a factor to bring the approximated two - photon rates of h ( equation ( 5 ) ) with @xmath0@xmath273 into agreement with the numerical value given above , i.e. @xmath63 note that the use of the non - resonant rates is an approximation . the resonant contributions are suppressed in practice because of optical depth effects , and in a sense some of these contributions are already included in our multi - level calculation . nevertheless , the correct way to treat these effects would be in a full radiative transfer calculation , which we leave for a future study . for hei , we treat @xmath45s@xmath37 and @xmath45d@xmath46 separately and use equation ( 3 ) for calculating the transition rates . the spontaneous rate @xmath64 is estimated by @xcite by assuming a similar form to that used for @xmath50 : @xmath65 where @xmath66 is a fitting parameter ( which is still uncertain both theoretically and experimentally ) . according to @xcite , resonable values of @xmath67 range from 10 to 12s@xmath54 , and we take @xmath67@xmath2711s@xmath54 here . in our calculation , we include these extra two - photon rates up to @xmath0@xmath2740 for h and up to @xmath68 for hei . the other additional channel included is the spin - forbidden transition between the triplet @xmath10p@xmath11 and singlet @xmath12s@xmath13 states in hei . this is an intercombination / semi - forbidden electric - dipole transition which emits a single photon and therefore we can calculate the corresponding net rate by using the bound - bound resonant rate expression , i.e. @xmath69 + where @xmath70 .\end{aligned}\ ] ] + here @xmath71 , @xmath72 and @xmath73 are the einstein coefficients , @xmath74 is the sobolev escape probability , @xmath75 is the sobolev optical depth ( see * ? ? ? * and references therein ) , @xmath76 is the wavelength of the energy difference between states @xmath77 and @xmath78 , and @xmath79 is the blackbody intensity with temperature @xmath29 . this @xmath10p@xmath11@xmath12s@xmath13 transition is not the lowest transition between the singlet and the triplet states . the lowest one is the forbidden magnetic - dipole transition between @xmath10s@xmath11 and @xmath12s@xmath13 , with einstein coefficient @xcite . however , this is much smaller than ( * ? ? ? * ; * ? ? ? * in preparation ) , so this transition can be neglected . note that @xcite used an older value of @xcite in their calculation . we use the bader - deuflhard semi - implicit numerical integration scheme ( see section 16.6 in * ? ? ? * ) to solve the above rate equations . all the numerical results are carried out using the @xmath80cdm model with cosmological parameters : @[email protected] ; @[email protected] ; @[email protected] ; @xmath84@xmath270 ; @[email protected] ; @[email protected] and @xmath87=0.73 ( consistent with those in * ? ? ? here @xmath85 is the primordial he abundance and @xmath86 the present background temperature . the recombination histories calculated using the previous multi - level code @xcite and the code in this paper are shown in fig . [ graphxe ] , where @xmath88 is the ionization fraction relative to hydrogen . as we have included more transitions in our model , and these give electrons more channels to cascade down to the ground state , we expect the overall recombination rate to speed up , and that this will be noticeable if the rates of the extra frobidden transitions are significant . from fig . [ graphxe ] , we can see that the recombination to hei is discernibly faster in the new calculation . [ gdxe ] shows the difference in @xmath4 with and without the extra forbidden transitions . the dip at around @xmath89 corresponds to the recombination of hei and the one around @xmath90 is for h. overall , the addition of the forbidden transitions claimed by @xcite leads to greater than 1 per cent change in @xmath4 over the redshift range where the cmb photons are last scattering . in the last section , we found that the approximated two - photon rate given by @xcite for h with @xmath0@xmath273 was overestimated by more than a factor of 10 . by considering only this extra two - photon transition , the approximate rate gives more than a per cent difference in @xmath4 , while with the more accurate numerical rates , the change in @xmath4 is less than 0.1 per cent ( as shown in fig . [ hn3 ] ) . based on this result , we do not need to include this two - photon transition , as it brings much less than a per cent effect on @xmath4 . for estimating the effect of the extra two - photon transitions for higher @xmath0 , we use the scaled two - photon rate given by equation ( 6 ) . the result is plotted in fig . [ gdxeh ] . the change in @xmath4 with the scaled two - photon rates is no more than 0.4 per cent , while the one with the @xcite approximated rates brings about a 5 per cent change . for hei , @xcite included the two - photon transitions from @xmath0@xmath276 to 40 , since they claimed that the approximate formula ( equation 7 ) is good for @xmath0@xmath916 . in our calculation , we use @xmath92 from the approximate formula for the two - photon transitions of @xmath0@xmath273 to 10 , since this is the best one can do for now ( and the formula at least gives the right order of magnitude ) . the addition of the singlet - triplet @xmath77@xmath78 transition and the @xmath93@xmath78 and @xmath94@xmath78 two - photon transitions with cause more than 1 per cent changes in @xmath4 ( as shown in fig . 2 ) . the @xmath77@xmath78 transition has the biggest effect on @xmath4 . [ gdxehei ] shows the fractional difference in @xmath4 using different combinations of additional forbidden transitions . we can see that the @xmath77@xmath78 transition alone causes more than a 1 per cent change in @xmath4 , and the addition of each two - photon transition only gives about another 0.1 per cent change . the extra two - photon transitions from higher excited states ( larger @xmath0 ) have a lower effect on @xmath4 compared with that from small @xmath0 , and we checked that this trend continues to higher @xmath0 . however , the convergence is slow with increasing @xmath0 . therefore , one should also consider these two - photon transitions with for hei , and the precise result will require the use of accurate rates , rather than an approximate formula such as equation ( 7 ) . for the @xmath77@xmath78 transition , @xcite adopted an older and slightly larger rate , and this causes a larger change of the ionization fraction ( about 0.5 per cent more compared with that calculated with our best rate ) , as shown in fig . [ gdxe23p ] . [ cols="^,^,^,^ " , ] one might wonder why the semi - forbidden transitions are significant in recombination _ at all _ , since the spontaneous rate ( or the einstein @xmath67 coefficient ) of the semi - forbidden transitions are about 6 orders of magnitude ( a factor of @xmath95 , where @xmath96 is the fine - sturcture constant ) smaller than those of the resonant transitions . let us take hei as an example for explaining the importance of the spin - forbidden @xmath77@xmath78 transition in recombination . the spontaneous rate is equal to 177.58s@xmath54 for this semi - forbidden transition , which is much smaller than @xmath97s@xmath54 for the @xmath98@xmath78 resonant transition . but when we calculate the net rate [ see equation ( 8) ] , we also need to include the effect of absorption of the emitted photons by the surrounding neutral atoms , and we take this into account by multiplying the net bound - bound rate by the sobolev escape probability @xmath99 @xcite . if @xmath99=1 , the emitted line photons can escape to infinity , while if @xmath99=0 the photons will all be reabsorbed and the line is optically thick . [ gescp ] shows that the escape probability of the @xmath98@xmath78 resonant transition is about 7 orders of magnitude smaller than the spin - forbidden transition . this makes the two net rates roughly comparable , as shown in fig . [ gratehei ] . from equation ( 11 ) , we can see that the easier it is to emit a photon , the easier that photon can be re - absorbed , because the optical depth @xmath75 is directly proportional to the einstein @xmath67 coefficient . so when radiative effects dominate , it is actually natural to expect that some forbidden transitions might be important ( although this is not true in a regime where collisonal rates dominate which is often the case in astrophysics ) . in fact for today s standard cosmological model , slightly more than half of all the hydrogen atoms in the universe recombined via a forbidden transition @xcite . table 1 shows that this is also true for helium . in the previous multi - level calculation @xcite , there was no direct transition between the singlet and triplet states . the only communication between them was via the continuum , through the photo - ionization and photo - recombination transitions . table 1 shows how many electrons cascade down through each channel from @xmath100 states to the ground state . in the previous calculation , about 70% of the electrons went down through the @xmath98@xmath78 resonant transition . in the new calculation , including the spin - forbidden transition between the triplets and singlets , there are approximately the same fraction of electrons going from the @xmath98 and @xmath77 states to the ground state ( actually slightly more going from @xmath10p@xmath11 in the current cosmological model ) . this shows that we should certainly include this forbidden transition in future calculations . our estimate is that only about 40% of helium atoms reach the ground state without going through a forbidden transition . how about the effect of other forbidden transitions in hei recombination ? we have included all the semi - forbidden electric - dipole transitions with @xmath0@xmath25@xmath101 and @xmath8@xmath25@xmath102 , and with oscillator strengths larger than @xmath103 given by @xcite ( private communication ) . there is no significant change found in the ionization fraction . besides the @xmath10p@xmath11@xmath78 transition , all the other extra semi - forbidden transitions are among the higher excited states where the resonant transitions dominate . this is because these transition lines are optically thin and the escape probabilities are close to 1 . the cmb anisotropy power spectrum @xmath104 depends on the detailed profile of the evolution of the ionization fraction @xmath4 . this determines the thickness of the photon last scattering surface , through the visibility function , where @xmath105 is the thomson scattering optical depth . the function @xmath106 sets the epoch when the tight coupling between baryons and photons breaks down , i.e. when the photon diffusion length becomes long , and the visibility function fixes the time when the fluctuations are effectively frozen in ( see * ? ? ? * ; * ? ? ? * and references therein ) . the addition of the extra forbidden transitions speeds up both the recombination of h and he i , and hence we expect that there will be changes in @xmath5 . in order to perform the required calculation , we have used the code cmbfast @xcite and modified it to allow the input of an arbitrary recombination history . [ gcltt ] and [ gclee ] show the relative changes in the cmb temperature ( _ tt _ ) and polarization ( _ ee _ ) anisotropy spectra , respectively , with different combinations of extra forbidden transitions . the overall decrease of free electrons brings a suppression of @xmath5 over a wide range of @xmath8 . for hei , there is less @xmath4 at @xmath16@xmath57@xmath107 , which leads to an ealier relaxation of tight coupling . therefore , both the photon mean free path and the diffusion length are longer . moreover , the decrease of @xmath4 in the high-@xmath16 tail results in increased damping , since the effective damping scale is an average over the visibility function . this larger damping scale leads to suppression of the high-@xmath8 part of the power spectrum . from figs . [ gcltt ] and [ gclee ] , we can see a decrease of @xmath5 ( for both _ tt _ and _ ee _ ) toward high @xmath8 for hei , with the maximum change being about 0.6 percent . for h , the change of @xmath5 is due to the decrease in @xmath4 at @xmath16@xmath57@xmath108 ( see fig . [ gdxe ] ) . there are two basic features in the curve of change in @xmath104 ( the dotted and dashed lines in fig . [ gcltt ] ) . firstly , the power spectrum is suppressed with increasing @xmath8 , due to the lower @xmath4 in the high-@xmath16 tail ( @xmath16@xmath911000 ) . secondly , there are a series of wiggles , showing that the locations of the acoustic peaks are slightly shifted . this is due to the change in the time of generation of the @xmath5s in the low-@xmath16 tail . @xmath109 actually shows an increase for @xmath110 ( see fig . [ gclee ] ) ; this is caused by the shift of the center of the visibility function to higher @xmath16 , leading to a longer diffusion length . polarization occurs when the anisotropic hot and cold photons are scattered by the electrons . the hot and cold photons can interact with each other within the diffusion length , and therefore , a longer diffusion length allows more scatterings and leads to a higher intensity of polarization at large scales . with the approximate rates used by @xcite , the maximum relative change of @xmath111 is about 4 percent and for @xmath109 it is about 6 percent . the overall change is thus more than 1 percent over a wide range of @xmath8 . however , if we adopt the scaled two - photon rate given by equation ( 6 ) , the relative changes of @xmath111 and @xmath109 are no more than 1 per cent . note that we do not plot the temperature - polarization correlation power spectrum here , since there is no dramatically different change found ( and relative differences are less meaningful since @xmath112 oscillates around zero ) . in our model we only consider the semi - forbidden transitions with @xmath0@xmath2510 and @xmath8@xmath257 for hei and the two - photon transitions from the higher s and d states to the ground state for h and hei . it would be desirable to perform a more detailed investigation of all the other forbidden transitions , which may provide more paths for the electrons to cascade down to the ground state and speed up the recombination process . in this paper we have tried to focus on the forbidden transitions which are likely to be the most significant . however we caution that , if the approximations used are inadequate , or other transitions prove to be important , then our results will not be accurate . there are several other approximations that we have adopted in order to perform our calculations . for example , we consider the _ non - resonant _ two - photon rates for higher excited rates . the two - photon transitions from higher excited states ( @xmath0@xmath563 ) to the ground state are more complicated than the 2s1s transition , because of the resonant intermediate states . for example , for the 3s1s two - photon transition , the spectral distribution of the emitted photons shows infinities ( resonance peaks ) at the frequencies corresponding to the 3s2p and 2p1s transitions @xcite . here , we use only the non - resonant rates , by assuming a smooth spectral distribution of the emitted photons ; this probably gives a lower limit on the change of @xmath4 and @xmath5 coming from these extra forbidden transitions . the correct way to treat this would be to consider the rates and feedback from medium using the full spectral distribution of the photons and radiative transfer ; this will have to wait for a future study . besides the consideration of more forbidden transitions , there are many other improvements that could be made to the recombination calculation . in particular , @xcite showed that a multi - level calculation of the recombination of h with the inclusion of seperate @xmath8-states can give more than 20 per cent difference in the population of some levels compared with the thermal equilibrium assumption for each @xmath0-shell . the latest calculation , considering up to 100 shells , is presented by @xcite , but does not include all the forbidden transitions studied here . a more complete calculation should be done by combining the forbidden transitions in a code with full angular momentum states , and we leave this to a future study . there are also other elaborations which could be included in future calculations , which we now describe briefly . the rate equation we use for all the two - photon transitions only includes the spontaneous term , assuming there is no interaction with the radiation background ( see equation ( 3 ) ) . @xcite suggested that one should also consider the stimulated effect of the 2s1s two - photon transition for h , due to photons in the low frequency tail of the cmb blackbody spectrum . @xcite additionally argued that the change of the adiabatic index of the matter should also be included , arising due to the neutralization of the ionized gas . these two modifications have been studied only in an effective three - level atom model , and more than a percent change in @xmath4 was claimed in each case ( but see @xcite for arguments against the effect claimed by leung et al . 2004 ) . for the background radiation field @xmath79 , we approximated it with a perfect blackbody planck spectrum . this approximation is not completely correct for the recombination of h , since the he line distortion photons redshift into a frequency range that can in principle photo - ionize the neutral h @xcite . althought we expect this secondary distortion effect to bring the smallest change on @xmath4 among all the modifications suggested here , it is nevertheless important to carry out the calculation self - consistently , particularly for the spectral line distortions . in order to obtain an accurate recombination history , we therefore need to perform a full multi - level calculation with seperate @xmath8-states and all the improvements suggested above , which we plan to do in a later paper . for completeness we also point out that the accuracy of the physical constants is important for recombination as well . the most uncertain physical quantity in the recombination calculation is the gravitational constant @xmath113 . the value of @xmath113 used previously in the recfast code is @xmath114@xmath115@xmath116m@xmath117kg@xmath54s@xmath118 and the latest value ( e.g. from the particle data group , * ? ? ? * ) is @xmath119@xmath115@xmath116m@xmath117kg@xmath54s@xmath118 . another quantity we need to modify is the atomic mass ratio of @xmath120he and @xmath121h , @xmath122 , which was previously taken to be equal to 4(as pointed out by * ? ? ? by using the atomic masses given by @xcite , the mass ratio is equal to 3.9715 . the overall change in @xmath4 is no more than 0.1 per cent after updating these two constants in both recfast and multi - level code . in this paper , we have computed the cosmological recombination history by using a multi - level code with the addition of the @xmath10p@xmath11 to @xmath12s@xmath13 spin - forbidden transition for hei and the two - photon transitions from @xmath0s and @xmath0d states to the ground state for both h and hei . with the approximate rates from @xcite , we find that there is more than a per cent decrease in the ionization fraction , which agrees broadly with the result they claimed . however , the only available accurate numerical value of two - photon rate with @xmath123 is for the 3s to 1s and 3d to 1s transitons for h. we found that the approximate rates from @xcite were overestimated , and instead we considered a scaled rate in order to agree with the numerical @xmath124 two - photon rate . with this scaled rate , the change in @xmath4 is no more than 0.5 per cent . including these extra forbidden transitions , the change in the cmb anisotropy power spectrum is more than 1 per cent , which will potentially affect the determination of cosmological parameters in future cmb experiments . since one would like the level of theoretical uncertainty to be negligible , it is essential to include these forbidden transitions in the recombination calculation . in addition , we still require accurate spontaneous rates to be calculated for the two - photon transitions and also a code which includes at least all the modifications suggested in section 4 , in order to obtain the @xmath5s down to the 1 per cent level . achieving sub - percent accuracy in the calculations is challenging ! however , the stakes are high the determination of the parameters which describe the entire universe and so further work will be necessary . systematic deviations of the sort we have shown would potentially lead to incorrect values for the spectral tilt derived from planck and even more ambitions future cmb experiments , and hence incorrect inferences about the physics which produced the density perturbations in the very early universe . it is amusing that in order to understand physics at the @xmath125gev energy scale we need to understand ev scale physics in exquisite detail ! this work would have been impossible without the earlier discussions with sara seager . we thank donald c. morton for useful conversations and providing us with updated forbidden transition rates . we are grateful to peter j. storey for providing the data for the photo - ionization cross - section of neutral helium . we would also like to thank alexander dalgarno , victor k. dubrovich , christopher hirata and eric switzer for useful discussions . this work made use of the code cmbfast , written by uros seljak and matias zaldarriaga . this research was supported by the natural sciences and engineering research council of canada and a university of british columbia graduate fellowship . 99 bauman r.p . , porter r.l . , ferland g.j . , macadam k.b . , 2005 , apj , 628 , 541 chluba j. , sunyaev r.a . , 2006 , a&a , 446 , 39 chluba j. , rubino - martin j.a . , sunyaev r.a . , 2006 , astro - ph/0608242 cresser j.d . , tang a.z . , salamo g.j . , chan f.t . , 1986 , phys . a , 33 , 3 , 1677 dellantonio i.p . , rybicki g.b . , 1993 , in asp conf . 51 , observational cosmology , ed . g. chincarini et al . ( san francisco : asp ) , 548 drake g.w.f . , morton d.c . , 2006 , in preparation for the astrophysical journal dubrovich v.k . , grachev s.i . , 2005 , astronomy letters , 31 , 6 , 35 florescu v. , patrascu s. , stoican o. , 1987 , phys . rev . a , 36 , 2155 florescu v. , schneider i. , mihailescu i.n . , 1988 , phys a , 38 , 4 , 2189 goldman s.p . , 1989 , phys . a , 40 , 1185 hu w. , scott d. , sugiyama n. , white m. , 1995 , phys . d , 52 , 5498 hummer d.g . , storey p.j . , 1998 , mnras , 297 , 1073 labzowsky l.n . , shonin a.v . , solovyev d.a . , 2005 , journal of physics b , 38 , 265 lach g. , pachucki k. , 2001 , phys . rev . a , 64 , 042510 leung p.k . , chan c.w . , chu m.c . , 2004 , mnras , 349 , 2 , 632 lewis a. , weller j. , battye r. , 2006 , astro - ph/0606552 lin c.d . , johnson w.r . , dalgarno a. , 1977 , phys . a , 15 , 1 , 154 morton d.c . , wu q. , drake g.w.f . , 2006 , can . j. phys . , 84 , 83 peebles p.j.e . 1968 , apj , 153 , 1 planck , 2006 , esa - sci(2005)1 press w.h . , flannery b.p . , teukolsky s.a . , vetterling w.t . , 1992 , numerical recipes in c : the art of scientific computing , cambridge univ . press , cambridge , uk rubino - martin j.a . , chluba j. , sunyaev r.a . , 2006 , astro - ph/0607373 santos j.p . , parente f. , indelicato p. , 1998 j. , d3 , 43 seager s. , sasselov d.d . , scott d. , 1999 , apj , 523 , l1 seager s. , sasselov d.d . , scott d. , 2000 , apjs , 128 , 407 seljak u. , sugiyama n. , white m. , zaldarriaga m. , 2003 , phys . rev . d , 68 , 083507 seljak u. , zaldarriaga m. , 1996 , apj , 463 , 1 spergel d.n . , el al . , 2006 , astro - ph/0603449 steigman g. , 2006 , astro - ph/0606206 storey p.j . , hummer d.g . , 1991 , comput commun . , 66 , 12 tung j.h . , ye x.m . , salamo g.j . , chan f.t . , 1984 , a , 30 , 1175 yao w .- m . , 2006 , journal of physics g , 33 , 1 wong w.y . , seager s. , scott d. , 2006 , mnras , 367 , 1666 wong w.y . , scottd . , 2006 , astro - ph/0612322 zeldovich y.b . , kurt v.g . , sunyaev r.a . , 1968 , zh fiz . , 55 , 278 ; english transl . , 1969 , soviet phys . jetp lett . , 28 , 146	more than half of the atoms in the universe recombined via forbidden transitions , so that accurate treatment of the forbidden channels is important in order to follow the cosmological recombination process with the level of precision required by future microwave anisotropy experiments . we perform a multi - level calculation of the recombination of hydrogen ( h ) and helium ( he ) with the addition of the spin - forbidden transition for neutral helium ( hei ) , plus the @xmath0s1s and @xmath0d1s two - photon transitions for h ( up to @xmath1 ) and among singlet states of hei ( @xmath2 and @xmath3 ) . the potential importance of such transitions was first proposed by @xcite using an effective three - level atom model . here , we relax the thermal equilibrium assumption among the higher excited states to investigate the effect of these extra forbidden transitions on the ionization fraction @xmath4 and the cosmic microwave background ( cmb ) angular power spectrum @xmath5 . the spin - forbidden transition brings more than a percent change in @xmath4 . the two - photon transitions may also give non - negligible effects , but currently accurate rates exist only for @xmath6 . we find that changes in both @xmath4 and @xmath5 would be at about the percent level with the approximate rates given by @xcite . however , the two - photon rates from 3s to 1s and 3d to 1s of h appear to have been overestimated , and our best numerical calculation puts the effect on @xmath4 and @xmath5 at below the percent level . nevertheless , we do not claim that we have the definite answer , since several issues remain open ; sub - percent level computation of the @xmath5s requires improved calculations of atomic transition rates as well as increasingly complex multi - level atom calculations . [ firstpage ] cosmology : cosmic microwave background cosmology : early universe cosmology : theory atomic processes .
the galaxy two - point correlation function , @xmath12 , is a fundamental statistic of the galaxy distribution , and is relatively straightforward to calculate from observational data . since the clustering of galaxies is determined by the initial mass fluctuations and their evolution , measurements of @xmath12 set constraints on the initial mass fluctuations and their evolution . the astrophysics of galaxy formation introduces uncertainties , but there is now good evidence that galaxies do trace the underlying mass distribution on large scales . in this paper we analyse the distribution of @xmath24 220000 galaxies in the 2-degree field galaxy redshift survey ( 2dfgrs , colless 2001 ) . a brief summary of the data is presented in section 2 . much of our error analysis makes use of mock galaxy catalogues generated from @xmath25-body simulations which are also discussed in section 2 . the two - dimensional measurement @xmath0 , where @xmath26 is the pair separation perpendicular to the line - of - sight and @xmath27 is the pair separation parallel to the line - of - sight provides information about the real - space correlation function , the small - scale velocity distribution , and the systematic gravitational infall into overdense regions . the spherical average of @xmath0 gives an estimate of the redshift - space correlation function , @xmath4 , where @xmath28 , so the galaxy separations are calculated assuming that redshift gives a direct measure of distance , ignoring the effects of peculiar velocities . integrating @xmath0 along the line - of - sight sums over any peculiar velocity distributions , and so is unaffected by any redshift - space effects . the resulting projected correlation function , @xmath6 , is directly related to the real - space correlation function . our estimates of @xmath0 , @xmath4 , @xmath6 and @xmath7 are presented in section 3 . these statistics have been measured from many smaller redshift surveys ( davis & peebles 1983 ; loveday 1992 ; jing , mo & brner 1998 ; hawkins 2001 ; zehavi 2002 ) , but since they sample smaller volumes , there is a large cosmic variance on the results . the large volume sampled by the 2dfgrs leads to significantly more reliable estimates . a preliminary analysis was performed on the 2dfgrs by peacock ( 2001 ) but we now have a far more uniform sample and twice as many galaxies . madgwick ( 2003 ) have measured these statistics for spectral - type sub - samples of the 2dfgrs . peculiar velocities of galaxies lead to systematic differences between redshift - space and real - space measurements , and we can consider the effects in terms of a combination of large - scale coherent flows induced by the gravity of large - scale structures , and a small - scale random peculiar velocity of each galaxy ( e.g. marzke 1995 ; jing 1998 ) . the large - scale flows compress the contours of @xmath0 along the @xmath27 direction , as described by kaiser ( 1987 ) and hamilton ( 1992 ) . the amplitude of the distortion depends on the mean density of the universe , @xmath29 , and on how the mass distribution is clustered relative to galaxies , which can be parameterized in terms of a linear bias @xmath30 , defined so that @xmath31 , where @xmath32 represents fluctuations in the density field . the random component of peculiar velocity for each galaxy means that the observed @xmath0 is convolved in the @xmath27 coordinate with the pairwise distribution of random velocities . in section 4 we describe the construction of a model @xmath0 from these assumptions about redshift - space distortions and also the shape of the correlation function . in section 5 we use the @xmath33 statistic ( hamilton 1992 ) based on the quadrupole moment of @xmath0 to estimate the parameter @xmath34 . in the absence of the small - scale random velocities the shape of @xmath0 contours on large scales is directly related to the parameter @xmath15 . a similar estimate of @xmath15 is provided by the ratio of amplitudes of @xmath4 to @xmath7 and this is also presented in section 5 . in section 6 , we use the landy , szalay & broadhurst ( 1998 ) method to estimate the distribution of peculiar velocities . this technique ignores the effect of large - scale distortions and uses the fourier transform of @xmath0 to estimate the distribution of peculiar velocities , @xmath17 . the large sample volume of the 2dfgrs makes our measurements more reliable than previous estimates in the same way as for the correlation functions mentioned earlier . these two approaches provide reasonable estimates of @xmath15 and @xmath17 so long as the distortions at small and large scales are completely decoupled . this is not the case for real data , and so we have fitted models which simultaneously include the effects of both @xmath15 and @xmath17 . the resulting best - fit parameters are the most self consistent estimates . previous data - sets have lacked the signal - to - noise to allow a reliable multi - parameter fit in this way . our fitting procedure and results are described in section 7 . in section 8 , we examine the luminosity and redshift dependence of @xmath15 and combine our results with estimates of @xmath30 ( verde 2002 ; lahav 2002 ) to estimate @xmath29 , and compare this with other recent analyses . in section 9 , we summarise our main conclusions . when converting from redshift to distance we assume the universe has a flat geometry with @xmath35 , @xmath36 and @xmath37 , so that all scales are in units of @xmath38 . the 2dfgrs is selected in the photometric @xmath39 band from the apm galaxy survey ( maddox , efstathiou & sutherland 1990 ) and its subsequent extensions ( maddox , in preparation ) . the bulk of the solid angle of the survey is made up of two broad strips , one in the south galactic pole region ( sgp ) covering approximately @xmath40 , @xmath41 and the other in the direction of the north galactic pole ( ngp ) , spanning @xmath42 , @xmath43 . in addition to these contiguous regions , there are a number of circular 2-degree fields scattered randomly over the full extent of the low extinction regions of the southern apm galaxy survey . the magnitude limit at the start of the survey was set at @xmath44 but both the photometry of the input catalogue and the dust extinction map have been revised since and so there are small variations in magnitude limit as a function of position over the sky . the effective median magnitude limit , over the area of the survey , is @xmath45 ( colless 2001 ) . the completeness of the survey data varies according to the position on the sky because of unobserved fields ( mostly around the survey edges ) , un - fibred objects in observed fields ( due to collision constraints or broken fibres ) and observed objects with poor spectra . the variation in completeness is mapped out using a completeness mask ( colless 2001 ; norberg 2002a ) which is shown in fig . [ f : masks ] for the data used in this paper . we use the data obtained prior to may 2002 , which is virtually the completed survey . this includes 221283 unique , reliable galaxy redshifts ( quality flag @xmath46 3 , colless 2001 ) . we analyse a magnitude - limited sample with redshift limits @xmath47 and @xmath48 , and no redshifts are used from a field with @xmath49 completeness . the median redshift is @xmath50 . the random fields , which contain nearly 25000 reliable redshifts are not included in this analysis . after the cuts for redshift , completeness and quality we are left with 165659 galaxies in total , 95929 in the sgp and 69730 in the ngp . these data cover an area , weighted by the completeness shown in fig . [ f : masks ] , of @xmath51 in the sgp and @xmath52 in the ngp , to the magnitude limit of the survey . in all of the following analysis we consider the ngp and sgp as independent data sets . treating the ngp and sgp as two independent regions of the sky gives two estimates for each statistic , and so provides a good test of the error bars we derive from mock catalogues ( see below ) . we have also combined the two measurements to produce our overall best estimate by simply adding the pair counts from the ngp and sgp . the optimal weighting of the two estimates depends on the relative volumes surveyed in the ngp and sgp , but since these are comparable , a simple sum is close to the optimal combination . it is important to estimate the effective redshift at which all our statistics are calculated . as @xmath12 is based on counting pairs of galaxies the effective redshift is not the median , but a pair - weighted measure . the tail of high redshift galaxies pushes this effective redshift to @xmath2 . similarly the effective magnitude of the sample we analyse is @xmath53 , corresponding to @xmath54 ( using @xmath55 , norberg 2002a ) . for each of the ngp and sgp regions , 22 mock catalogues were generated from the @xmath13 hubble volume simulation ( evrard 2002 ) using the techniques described in cole ( 1998 ) , and are designed to have a similar clustering signal to the 2dfgrs . a summary of the construction methods is presented here but for more details see norberg ( 2002a ) and baugh ( in preparation ) . these simulations used an initial dark matter power spectrum appropriate to a flat @xmath13 model with @xmath36 and @xmath56 . the dark - matter evolution was followed up to the present day and then a bias scheme ( model 2 of cole 1998 , with a smoothing length , @xmath57 ) was used to identify galaxies from the dark matter haloes . the bias scheme used has two free parameters which are adjusted to match the mean slope and amplitude of the correlation function on scales greater than a few megaparsec . on scales smaller than the smoothing length there is little control over the form of the clustering , but in reality the methods employed work reasonably well ( see later sections ) . the resulting catalogues have a bias scheme which asymptotes to a constant on large scales , giving @xmath58 , but is scale dependent on small scales . apparent magnitudes were assigned to the galaxies consistent with their redshift , the assumed schechter luminosity function and the magnitude limit of the survey . the schechter function has essentially the same parameters as in the real data ( see norberg 2002a ) . then the completeness mask and variable apparent magnitude limits were applied to the mock catalogues to reproduce catalogues similar to the real data . in the analysis which follows , we make use of the real- and redshift - space correlation functions from the full hubble volume simulation . these correlation functions are determined from a fourier transform of the power spectrum of the full hubble volume cube using the real- and redshift - space positions of the mass particles respectively , along with the bias scheme outlined above . this allows us to compare our mock catalogue results with that of the simulation from which they are drawn to ensure we can reproduce the correct parameters . it also allows us to compare and contrast the results from the real universe with a large numerical simulation . we analyse each of the mock catalogues in the same way as the real data , so that we have 22 mock measurements for every measurement that we make on the real data . the standard deviation between the 22 mock measurements gives a robust estimate of the uncertainty on the real data . we use this approach to estimate the uncertainties for direct measurements from the data , such as the individual points in the correlation function , and for best - fit parameters such as @xmath59 . when fitting parameters we use this standard deviation as a weight for each data - point and perform a minimum @xmath60 analysis to obtain the best - fit parameter . the errors that we quote for any particular parameter are the rms spread between the 22 best - fit parameters obtained in the same way from the mock catalogues . this simple way of estimating the uncertainties avoids the complications of dealing directly with correlated errors in measured data points , while still providing an unbiased estimate of the real uncertainties in the data , including the effects of correlated errors . although this approach gives reliable estimates of the uncertainties , the simple weighting scheme is not necessarily optimal in the presence of correlated errors . nevertheless , for all statistics that we consider , we find that the means of the mock estimates agree well with the values input to the parent simulations . also , we have applied the technique described by madgwick ( 2003 ) to de - correlate the errors for the projected correlation function , using the covariance matrix estimated from the mock catalogues . we found a @xmath61 difference between the best - fit values using the two methods . so , we are confident that our measurements and uncertainty estimates are robust and unbiased . the two point correlation function , @xmath12 , is measured by comparing the actual galaxy distribution to a catalogue of randomly distributed galaxies . these randomly distributed galaxies are subject to the same redshift , magnitude and mask constraints as the real data and we modulate the surface density of points in the random catalogue to follow the completeness variations . we count the pairs in bins of separation along the line - of - sight , @xmath27 , and across the line - of - sight , @xmath26 , to estimate @xmath0 . spherically averaging these pair counts provides the redshift - space correlation function @xmath4 . finally , we estimate the projected function @xmath6 by integrating over all velocity separations along the line - of - sight and invert it to obtain @xmath7 . to reduce shot noise we compare the data with a random catalogue containing ten times as many points as the real catalogue . this random catalogue needs to have a smooth selection function matching the @xmath62 of the real data . we use the 2dfgrs luminosity function ( norberg 2002a ) with @xmath63 and @xmath64 to generate the selection function , following the change in the survey magnitude limit across the sky . when analysing the mock catalogues , we use the input luminosity function to generate the selection function , and hence random catalogues . as an alternative method , we also fitted an analytic form for the selection function ( baugh & efstathiou 1993 ) to the data , and generated random catalogues using that selection function . we have calculated all of our statistics using both approaches , and found that they gave essentially identical results for the data . when analysing the mock catalogues , we found that the luminosity function method was more robust to the presence of large - scale features in the @xmath62 data . thus , all of our quoted results are based on random catalogues generated using the luminosity function . the design of the 2df instrument means that fibres can not be placed closer than approximately 30 arcsec ( lewis 2002 ) , and so both members of a close pair of galaxies can not be targeted in a single fibre configuration . fortunately , the arrangement of 2dfgrs tiles means that not all close pairs are lost from the survey . neighbouring tiles have significant areas of overlap , and so much of the sky is targeted more than once . this allows us to target both galaxies in some close pairs . nevertheless , the survey misses a large fraction of close pairs . it is important to assess the impact of this omission on the measurement of galaxy clustering and to investigate schemes that can compensate for the loss of close pairs . to quantify the effect of these so - called ` fibre collisions ' we have calculated the angular correlation function for galaxies in the 2dfgrs parent catalogue , @xmath65 , and for galaxies with redshifts used in our @xmath12 analysis , @xmath66 . we used the same mask to determine the angular selection and apparent magnitude limit for each sample as in fig . [ f : masks ] . note that the mask is used only to define the area of analysis , and the actual redshift completeness values are not used in the calculation of @xmath67 . in our @xmath12 analyses we impose redshift limits @xmath68 , which means that the mean redshift of the redshift sample is lower than the parent sample . we used limber s equation ( limber 1954 ) to calculate the scale factors in amplitude and angular scale needed to account for the different redshift distributions . the solid line in fig . [ f : ang ] shows @xmath69 , and the filled circles show @xmath70 after applying the limber scale factors . the error bars in fig . [ f : ang ] show @xmath71 from the full apm survey ( maddox , efstathiou & sutherland 1996 ) , also scaled to the magnitude limit of the 2dfgrs parent sample . on scales @xmath72 all three measurements are consistent . on smaller scales @xmath70 is clearly much lower than @xmath69 , showing that the fibre collision effect becomes significant and can not be neglected . the ratio of galaxy pairs counted in the parent and redshift samples is given by @xmath73 , which is shown by the filled circles in the lower panel of fig . [ f : ang ] . as discussed in the next section , we use this ratio to correct the pair counts in the @xmath12 analysis . each galaxy and random galaxy is given a weighting factor depending on its redshift and position on the sky . the redshift dependent part of the weight is designed to minimize the variance on the estimated @xmath12 ( efstathiou 1988 ; loveday 1995 ) , and is given by @xmath74 , where @xmath75 is the density distribution and @xmath76 . we use @xmath75 from the random catalogue to ensure that the weights vary smoothly with redshift . we find that our results are insensitive to the precise form of @xmath77 but we derived it using a power law @xmath12 with @xmath78 and @xmath79 and a maximum value of @xmath80 . this corresponds to the best - fit power law over the range @xmath81 with a cutoff at larger scales . we also use the weighting scheme to correct for the galaxies that are not observed due to the fibre collisions . each galaxy - galaxy pair is weighted by the ratio @xmath82 at the relevant angular separation according to the curve plotted in the bottom panel of fig.[f : ang ] . this corrects the observed pair count to what would have been counted in the parent catalogue . the open points in fig . [ f : ang ] , which have the collision correction applied , show that this method can correctly recover the parent catalogue result and hence overcome the fibre collision problem . since the random catalogues do not have any close - pair constraints , only the galaxy - galaxy pair count needs correcting in this way . we also tried an alternative approach to the fibre - collision correction that we used previously in norberg ( 2001 , 2002b ) where the weight for each unobserved galaxy was assigned equally to its ten nearest neighbours . this produced similar results for @xmath83 , but did not help on smaller scales . all of our results are presented using the @xmath84 weighting scheme . hence each galaxy , @xmath85 , is weighted by the factor , @xmath86 and each galaxy - galaxy pair @xmath85,@xmath87 is given a weight @xmath88 , whereas each galaxy - random and random - random pair is given a weight @xmath89 . we use the @xmath12 estimator of landy & szalay ( 1993 ) , @xmath90 where @xmath91 is the normalised sum of weights of galaxy - galaxy pairs with particular @xmath92 separation , @xmath93 the normalised sum of weights of random - random pairs with the same separation in the random catalogue and @xmath94 the normalised sum of weights of galaxy - random pairs with the same separation . to normalise the pair counts we ensure that the sum of weights of the random catalogue equal the sum of weights of the real galaxy catalogue , as a function of scale . we find that other estimators ( hamilton 1993 ) give similar results . the @xmath62 distributions for the data and random catalogues ( scaled so that the area under the curve is the same as for the observed data ) are shown in fig . [ f : nzs ] . it is clear that @xmath62 for the random catalogues are a reasonably smooth fit to @xmath62 for the data . norberg ( 2002a ) showed that large ` spikes ' in the @xmath62 are common in the mock catalogues , and so similar features in the data redshift distributions indicate normal structure . the resulting estimates of @xmath0 calculated separately for the sgp and ngp catalogues are shown in fig . [ f : sigpi ] , along with the combined result . the velocity distortions are clear at both small and large scales , and the signal - to - noise ratio is in general very high for @xmath26 and @xmath27 values less than @xmath95 ; it is @xmath96 in each @xmath97 bin at @xmath98 . at very large separations @xmath0 becomes very close to zero , showing no evidence for features that could be attributed to systematic photometric errors . we used an earlier version of the 2dfgrs catalogue to carry out a less detailed analysis of @xmath0 ( peacock 2001 ) . the current redshift sample has about 1.4 times as many galaxies , though more importantly it is more contiguous , and the revised photometry has improved the uniformity of the sample . nevertheless our new results are very similar to our earlier analysis , demonstrating the robustness of our results . the current larger sample allows us to trace @xmath12 out to larger scales with smaller uncertainties . also , in our present analysis we analyse mock catalogues to obtain error estimates which are more precise than the previous error approximation ( see section[s : john ] ) . averaging @xmath0 at constant @xmath99 gives the redshift - space correlation function , and our results for the ngp and sgp are plotted in fig.[f : redxi ] on both log and linear scales . the ngp and sgp measurements differ by about @xmath100 between @xmath101 and @xmath102 , and we find one mock whose ngp and sgp measurements disagree by this much , and so it is probably not significant . we tried shifting @xmath103 by @xmath104mag to better fit the @xmath62 at @xmath105 in the sgp , and this moved the data points by @xmath106 for @xmath107 . the redshift - space correlation function for the combined data is plotted in fig . [ f : redxi2 ] in the top panel . it is clear that the measured @xmath4 is not at all well represented by a universal power law on all scales , but we do make an estimate of the true value of the redshift - space correlation length , @xmath59 , by fitting a localised power - law of the form , @xmath108 using a least - squares fit to @xmath109 as a function of @xmath110 , using two points either side of @xmath111 . this also gives a value for the local redshift - space slope , @xmath112 . the best - fit parameters for the separate poles and combined estimates are listed in table [ t : bf ] . in the inset of fig . [ f : redxi2 ] we can see , at a low amplitude , that @xmath113 goes negative between @xmath114 . in the bottom panel of fig . [ f : redxi2 ] we examine the shape of @xmath4 more carefully . the points are the data divided by a small scale power law fitted on scales @xmath115 ( dashed line ) . the data are remarkably close to the power - law fit for this limited range of scales , and follow a smooth break towards zero for @xmath116 . the measurements from the hubble volume simulation are shown by the solid line , and it matches the data extremely well on scales @xmath117 . on smaller scales , where the algorithm for placing galaxies in the simulation has little control over the clustering amplitude ( as discussed in section [ s : mocks ] ) , there are discrepancies of order @xmath118 . the mean @xmath4 determined from the mock catalogues agrees well with the true redshift - space correlation function from the full hubble volume . this provides a good check that our weighting scheme and random catalogues have not introduced any biases in the analysis . redshift - space correlation functions have been measured from many redshift surveys , but direct comparisons between different surveys are not straightforward because galaxy clustering depends on the spectral type and luminosity of galaxies ( guzzo 2000 ; norberg 2002b ; madgwick 2003 ) . direct comparisons can be made only between surveys that are based on similar galaxy selection criteria . the 2dfgrs is selected using pseudo - total magnitudes in the @xmath39 band , and the three most similar surveys are the stromlo - apm survey ( sapm , loveday 1992 ) , the durham ukst survey ( ratcliffe 1998 ) and the eso slice project ( esp , guzzo 2000 ) . the las campanas redshift survey ( lcrs , lin 1996 , jing 1998 ) and sloan digital sky survey ( sdss , zehavi 2002 ) are selected in the @xmath119 band , but have a very large number of galaxies , and so are also interesting for comparisons . the non - power - law shape of @xmath4 makes it difficult to compare different measurements of @xmath59 and @xmath112 , because the values depend sensitively on the range of @xmath99 used in the fitting procedure . in fig . [ f : xis_comp](a ) we compare the @xmath4 measurements directly for the 2dfgrs , sapm , durham ukst and esp surveys . our estimate of @xmath4 is close to the mean of previous measurements , but the uncertainties are much smaller . although we quote uncertainties that are similar in size to previous measurements , we have used the scatter between mock catalogues to estimate them , rather than the poisson or boot - strap estimates that have been used before and which seriously underestimate the true uncertainties . [ f : xis_comp](b ) shows the 2dfgrs measurements together with the lcrs and sdss measurements . on scales @xmath120 there appears to be no significant differences between the surveys , but for @xmath121 the lcrs and sdss have a higher amplitude than the 2dfgrs . this difference is likely to be caused by the different galaxy selection for the surveys , though the sdss results shown are for the early data release ( edr ) and have larger errors than the 2dfgrs points . the 2dfgrs is selected using @xmath39 , whereas the sdss and lcrs are selected in red bands . since the red ( early type ) galaxies are more strongly clustered than blue ( late type ) galaxies ( zehavi 2002 ; and via spectral type , norberg 2002b ) , we should expect that @xmath12 will be higher for red selected surveys than a blue selected survey . this issue is examined further in madgwick ( 2003 ) . the redshift - space correlation function differs significantly from the real - space correlation function because of redshift - space distortions ( see section 4 ) . we can estimate the real - space correlation length , @xmath122 , by first calculating the projected correlation function , @xmath6 . this is related to @xmath0 via the equation , @xmath123 though in practice we set the upper limit in this integral to @xmath124 . the result is insensitive to this choice for @xmath125 for our data . since redshift space distortions move galaxy pairs only in the @xmath27 direction , and the integral represents a sum of pairs over all @xmath27 values , @xmath6 is independent of redshift - space distortions . it is simple to show that @xmath6 is directly related to the real - space correlation function ( davis & peebles 1983 ) , @xmath126 if the real - space correlation function is a power law this can be integrated analytically . we write @xmath127 , where the @xmath128 superscripts refer to the ` projected ' values , rather than the ` inverted ' values which are calculated in section [ s : inv ] and denoted by @xmath129 . with this notation we obtain , @xmath130 the parameters @xmath131 and @xmath132 can then be estimated from the measured @xmath6 , giving an estimate of the real - space clustering independent of any peculiar motions . the projected correlation functions for the ngp and sgp are shown in fig . [ f : xi ] and the combined data result is shown in fig.[f : xi2 ] . the best - fit values of @xmath131 and @xmath132 for @xmath133 are shown in table [ t : bf ] . over this range @xmath134 is an accurate power law , but it steepens for @xmath135 . this deviation from power - law behaviour limits the scales that can be probed using this approach . it is possible to estimate @xmath7 by directly inverting @xmath6 without making the assumption that it is a power law ( saunders , rowan - robinson & lawrence 1992 , hereafter s92 ) . they recast eqn.[e : proj ] into the form , @xmath136 assuming a step function for @xmath137 in bins centered on @xmath138 , and interpolating between values , @xmath139 for @xmath140 . s92 suggest that their method is only good for scales @xmath141 in the qdot survey because @xmath142 becomes comparable to the maximum scale out to which they can estimate @xmath143 . we can test the reliability of our inversion of the 2dfgrs data using the mock catalogues . in fig . [ f : xir_mock ] we show the mean @xmath7 as determined from the mock catalogues using the method of s92 . we compare this to the real - space correlation function determined directly from the hubble volume simulation , from which the mock catalogues are drawn . the agreement is excellent and shows that the method works and that we can recover the real - space correlation function out to @xmath144 . like s92 we find that beyond this scale the method begins to fail and the true @xmath7 is not recovered . we have applied this technique to the combined 2dfgrs data and obtain the real - space correlation function shown in fig . [ f : xir ] . the data are plotted out to only @xmath144 due to the limitations in the method described above . on small scales @xmath7 is well represented by a power law , and a best - fit over the range @xmath145 gives the results for @xmath146 and @xmath147 shown in table [ t : bf ] . the points in the bottom panel of fig . [ f : xir ] show the 2dfgrs data divided by the best - fit power law . it can be seen that at scales @xmath148 the data @xmath7 is close to the best - fit power - law but does show hints of non power - law behaviour ( see also discussion below ) . .best - fit parameters to @xmath12 . for @xmath59 and @xmath112 the fit to @xmath4 uses only points around @xmath149 . for @xmath132 , @xmath131 and a(@xmath131 ) the fit to @xmath134 uses all points with @xmath150 . for @xmath146 and @xmath147 the fit to the inverted @xmath7 uses all points with @xmath145 . in each case the errors quoted are the rms spread in the results obtained from the same analysis with the mock catalogues . [ cols="<,<,<,<",options="header " , ] table [ t : bf_comp ] lists @xmath122 and @xmath151 for the 2dfgrs and other surveys estimated using power - law fits to the projected correlation function @xmath6 . as mentioned in section[s : red ] the sapm , durham ukst and the esp are @xmath39 selected surveys , and so should be directly comparable to the 2dfgrs . the values of @xmath122 and @xmath151 for these surveys all agree to within one standard deviation , except @xmath122 for the esp , which appears to be significantly lower . it is likely that the quoted uncertainties for the esp and durham ukst parameters are underestimated since they did not include the effect of cosmic variance . since they each sample relatively small volumes , this will be a large effect . the sparse sampling strategy used in the sapm means that it has a large effective volume , and so the cosmic variance is small . as in section [ s : red ] , the red - selected surveys , lcrs and sdss , are significantly different from the other surveys . the discrepancies are most likely due to the fact that the amplitude of galaxy clustering depends on galaxy type , and that red - selected surveys have a different mix of galaxy types . we can make a very rough approximation of the expected change in @xmath12 by considering how the mean colour difference of early and late populations changes the relative fraction of the two populations when a magnitude limited sample is selected in different pass bands . zehavi ( 2002 ) split their @xmath152selected sdss sample into 19603 early - type galaxies and 9532 late - type galaxies . the mean @xmath153 colours are 0.5 and 0.9 respectively . the 2dfgrs is selected using @xmath39 which is close to @xmath154 , and so compared to the @xmath142 selection , the median depth for blue galaxies will be larger that for red galaxies . the number of early and late types will roughly scale in proportion to the volumes sampled , and so the ratio of early - to - late galaxies in the 2dfgrs will be roughly @xmath155 . note that this colour split leads to a very different ratio of early - to - late galaxies compared to the @xmath156 split used by madgwick ( 2003 ) . assuming the early and late correlation functions trace the same underlying field , the combined correlation function will be @xmath157 from the power law fits of zehavi , the ratio of bias values at 1 mpc is @xmath158 . inserting the different ratios @xmath159 appropriate to the red and blue selected samples we find that the expected ratio of @xmath12 for a red selected sample compared to a blue selected sample is roughly 1.36 . scaling the 2dfgrs values of @xmath160 and @xmath161 leads to a sdss value of @xmath162 for @xmath163 , within @xmath164 of the actual sdss value . this simple argument indicates that the observed difference in @xmath12 between the red and blue selected surveys is consistent with the different population mixes expected in the surveys . the extra surface brightness selection applied to the lcrs may also introduce significant biases . each survey is also likely to have a different effective luminosity and , as has been shown by norberg ( 2001 ) , this will cause clustering measurements to differ . the relation for 2dfgrs galaxies found by norberg ( 2001 ) was , @xmath165 which gives , for @xmath166 ( see section [ s:2dfdata ] ) , @xmath167 , which will allow direct comparisons with other surveys . when analysing redshift surveys it must be remembered that the distance to each galaxy is estimated from its redshift and is not the true distance . each galaxy has , superimposed on its hubble motion , a peculiar velocity due to the gravitational potential in its local environment . these peculiar velocities can be in any direction and , since this effect distorts the correlation function , it can be used to measure two important parameters . the peculiar velocities are caused by two effects . on small scales , random motions of the galaxies within groups cause a radial smearing known as the ` finger of god ' . on large scales gravitational instability leads to coherent infall into overdense regions and outflow from underdense regions . we analyse the observed redshift - space distortions by modeling @xmath0 . we start with a model of the real - space correlation function , @xmath7 , and include the effects of large - scale coherent infall , which is parameterized by @xmath34 , where @xmath30 is the linear bias parameter . we then convolve this with the form of the random pairwise motions . kaiser ( 1987 ) pointed out that , in the linear regime , the coherent infall velocities take a simple form in fourier space . hamilton ( 1992 ) translated these results into real space , @xmath168 where p@xmath169 are legendre polynomials , @xmath170 and @xmath171 is the angle between @xmath142 and @xmath27 . the relations between @xmath172 , @xmath7 and @xmath15 for a simple power - law @xmath173 are ( hamilton 1992 ) , @xmath174 @xmath175 @xmath176 the appendix has more details of this derivation and gives the equations for the case of non - power law forms of @xmath12 . we use these relations to create a model @xmath177 which we then convolve with the distribution function of random pairwise motions , @xmath17 , to give the final model @xmath0 ( peebles 1980 ) : @xmath178 and we choose to represent the random motions by an exponential form , @xmath179 where @xmath18 is the pairwise peculiar velocity dispersion ( often known as @xmath180 ) . an exponential form for the random motions has been found to fit the observed data better than other functional forms ( ratcliffe 1998 ; landy 2002 ; see also section [ s : expvel ] ) . in this model we make several assumptions . firstly , we assume a power - law for the correlation function . the power - law approximation is a good fit on scales @xmath181 but is not so good at larger scales . this limits the scales which we can probe using this method . in section [ s : gridfit ] , we consider non - power - law models for @xmath7 , and recalculate eqns . [ e : xi0 ] to [ e : xi4 ] using numerical integrals ( see appendix ) , allowing us to reliably use scales @xmath182 . secondly , we assume that the linear theory model described above holds on scales @xmath183 , which is almost certainly not true . we also consider this in section 7 . finally , we assume an exponential distribution of peculiar velocities with a constant velocity dispersion , @xmath18 , ( eqn . [ e : fv ] ) and this is discussed and justified in section [ s : expvel ] and section[s : pecvels ] . to illustrate the effect of redshift - space distortions on the @xmath0 plot we show four model @xmath0 s in fig . [ f : modsp ] . if there were no distortions , then the contours shown would be circular , as in the top left panel due to the isotropy of the real - space correlation function . on small @xmath26 scales the random peculiar velocities cause an elongation of the contours in the @xmath27 direction ( the bottom left panel ) . on larger scales there is the flattening of the contours ( top right panel ) due to the coherent infall . the bottom right panel is a model with both distortion effects included . comparing the models of @xmath0 to the 2dfgrs measurements in fig . [ f : sigpi ] it is clear that the data show the two distortion effects included in the models . in section [ s : gridfit ] we use the data to constrain the model directly , and deduce the best - fit model parameters . before using the model described above to measure the parameters simultaneously , we first use methods that have been used in previous studies . this allows a direct comparison between our results and previous work . the ratio of the redshift - space correlation function , @xmath4 , to the real - space correlation function , @xmath7 , in the linear regime gives an estimate of the redshift distortion parameter , @xmath15 ( see eqn . [ e : xi0 ] ) , @xmath184 our results for the combined 2dfgrs data , using the inverted form of @xmath7 , are shown in fig . [ f : ratios ] by the solid points . the mean of the mock catalogue results is shown by the white line , with the rms errors shaded and the estimate from the hubble volume is shown by the solid line . the data are consistent with a constant value , and hence linear theory , on scales @xmath185 . the mock catalogues and hubble volume results asymptote to @xmath186 , the true value of @xmath15 in the mocks . the 2dfgrs data in the range @xmath14 are best - fit by a ratio of @xmath187 , corresponding to @xmath188 . the maximum scale that we can use in this analysis is determined by the uncertainty on @xmath7 from the saunders inversion method discussed in section [ s : inv ] . we now measure @xmath15 using the quadrupole moment of the correlation function ( hamilton 1992 ) , @xmath189 where @xmath190 is given by , @xmath191 these equations assume the random peculiar velocities are negligible and hence measuring @xmath33 gives an estimate of @xmath15 . the random uncertainties in this method are small enough that we obtain reliable estimates on scales @xmath192 , as shown by the mock catalogues ( see below ) , but the data are noisy beyond these scales . [ f : q ] shows @xmath33 estimates for the combined 2dfgrs data with the inset showing the ngp and sgp separately . the effect of the random peculiar velocities can be clearly seen at small scales , causing @xmath33 to be negative . the best - fit value to the combined data for @xmath193 is @xmath194 , which gives a value for @xmath195 , where the error is from the rms spread in the mock catalogue results . the solid line represents a model with @xmath196 and @xmath197 , which matches the data well ( see section[s : fits ] ) . although asymptoting to a constant , the value of @xmath33 in the model is still increasing at @xmath198 . this shows that non - linear effects do introduce a small systematic error even at these scales though this bias is small compared to the random error . to check whether this method can correctly determine @xmath15 we use the mock catalogues . the data points in fig . [ f : mock_q ] are the mean values of @xmath33 from the mock catalogues , with error bars on the mean , and the dashed line is the true value of @xmath58 . the data points seem to converge on large scales to the correct value of @xmath33 . fitting to each mock catalogue in turn for @xmath16 gives a mean @xmath199 , corresponding to @xmath200 . as the models showed , the random velocities will lead to an underestimate of @xmath15 even at @xmath198 , causing the difference between the measured and true values . this all shows that we can determine @xmath15 with a slight bias but the error bars are large compared to the bias . the @xmath33 estimates from the individual mock catalogues show a high degree of correlation between points on varying scales and so the overall uncertainty in @xmath33 from averaging over all scales @xmath201 is not much smaller than the uncertainty from a single point . it is this fact which makes the spread in results from the mock catalogues vital in the estimation of the errors on our result ( also see section[s : john ] ) . to this point we have assumed that the random peculiar velocity distribution has an exponential form ( eqn . [ e : fv ] ) . this form has been used by many authors in the past and has been found to fit the data better than other forms ( e.g. ratcliffe 1998 ) . we test this for the 2dfgrs data by following a method similar to that of landy , szalay and broadhurst ( 1998 , hereafter lsb98 ) . to extract the peculiar velocity distribution , we need to deconvolve the real - space correlation function from the peculiar velocity distribution . we first take the 2-d fourier transform of the @xmath0 grid to give @xmath202 and then take cuts along the @xmath203 and @xmath204 axes which we denote by @xmath205 and @xmath206 respectively , so @xmath207 and @xmath208 . by the slicing - projection theorem ( see lsb98 ) these cuts are equivalent to the fourier transforms of the real - space projections of @xmath0 onto the @xmath26 and @xmath27 axes . the projection of @xmath0 onto the @xmath26 axis is a distortion free measurement of @xmath143 but the projection onto the @xmath27 axis gives us @xmath143 convolved with the peculiar velocity distribution , ignoring the effects of large - scale bulk flows . since a convolution in real space is a multiplication in fourier space , the ratio of @xmath205 to @xmath206 is the fourier transform , @xmath209 $ ] , of the velocity distribution that we want to estimate . all that is left is to inverse fourier transform this ratio to obtain the peculiar velocity distribution , @xmath17 . lsb98 cut their dataset at @xmath210 and applied a hann smoothing window ; we use all the raw data . landy ( 2002 , hereafter l02 ) used the lsb98 method on the 100k 2dfgrs public release data and his results are discussed below . fitting an exponential to the resulting @xmath17 curve gives a value for @xmath18 assuming that the infall contribution to the velocity distribution is negligible . lsb98 and l02 claim that their method is not sensitive to the infall velocities . we show here that this is not the case . the additional structure in the fourier transform of the velocity distribution found by l02 is a direct consequence of the infall velocities . to test the lsb98 method we apply the technique to our models , described in section [ s : dis ] , with and without a @xmath211 infall factor , using various scales , and with and without a hann window . in fig . [ f : fft ] we show the fourier transform of the peculiar velocity distribution and in fig . [ f : fv ] we show the distribution function itself . it is clear from fig . [ f : fft ] that the shape of the fourier transform at small @xmath212 is quite badly distorted by the infall velocities . this leads to a systematic error in the actual velocity distribution as seen in fig . [ f : fv ] , where the measured peculiar velocity dispersions are biased low , especially in the case where a smoothing window and a limited range of scales are used . in particular the peak of the fourier transform is not at @xmath213 , and the inferred @xmath17 goes negative for a range of velocities ( dashed lines in the lower panels of fig . [ f : fv ] ) . this clearly can not be interpreted as a physical velocity distribution ; the method infers negative values because the input model @xmath0 is not consistent with the initial assumption of the method , which is that all of the distortion in @xmath0 is due to random peculiar velocities . we conclude that both types of peculiar velocity need to be considered when making these measurements , and so our preferred results come from directly fitting to @xmath0 . a further complication with the real data is that @xmath17 may depend on the pair separation ( see discussion in section 7.3 ) . the solid line in fig . [ f : vfv ] shows @xmath214 $ ] for a model where @xmath18 varies from 500@xmath215 at @xmath216 to 300@xmath215 at @xmath217 . this is compared to a model with @xmath218 ( dashed line ) , a model with @xmath219 ( dotted line ) and a model where @xmath18 varies from 300@xmath215 at @xmath220 to @xmath221 at @xmath222 ( dot - dashed line ) . the models with varying @xmath18 are very close to their respective constant @xmath18 models at all @xmath212 values , showing that this method leads to an estimate of @xmath214 $ ] determined mainly by the value of @xmath18 at small @xmath26 . the mean of the peculiar velocity distributions for the mock catalogues is shown in fig . [ f : mveld ] . the distribution is compared to a model , shown as the solid line , with @xmath58 and an exponential @xmath17 , with dispersion , @xmath223 . the exact form of the peculiar velocities in the hubble volume , and hence mock catalogues , is not explicitly specified and it should not be expected to conform to this model exactly . the fourier transform of the peculiar velocity distribution for the combined 2dfgrs data are shown in fig . [ f : ffta ] compared to a best - fit model with @xmath224 and @xmath225 . [ f : velda ] shows the peculiar velocity distribution itself compared to the same model . we showed in section [ s : vfv ] ( with fig . [ f : vfv ] ) that this was likely to be the value of @xmath18 at small @xmath26 . the distribution of random pairwise velocities does appear to have an exponential form , with a @xmath15 influence . sheth ( 1996 ) and diaferio & geller ( 1996 ) have shown that an exponential peculiar velocity distribution is a result of gravitational processes . ignoring the infall l02 found @xmath226 , using the smaller , publicly available , sample of 2dfgrs galaxies . we made the same approximations and repeated his procedure on our larger sample , and find @xmath227 . using our data grid out to @xmath228 , with no smoothing and ignoring @xmath15 , gives @xmath229 . we have shown that the result in l02 is biased low by ignoring @xmath15 and that the infall must be properly considered in these analyses . as shown in fig . [ f : velda ] , our data are reasonably well described by an exponential model with @xmath196 and @xmath230 . we now fit our @xmath0 data grid to the models described in section [ s : dis ] , assuming a power - law form for the real - space correlation function . this model has four free parameters , @xmath15 , @xmath122 , @xmath151 and @xmath18 . the fits to the data are done by minimising @xmath231_{\rm{model } } - \log[1+\xi]_{\rm{data}}}{\log[1+\xi+\delta\xi]_{\rm{data}}-\log[1+\xi-\delta\xi]_{\rm{data}}}\right)^2,\ ] ] for @xmath232 , where @xmath233 is the rms of @xmath12 from the mock catalogues for a particular @xmath26 and @xmath27 . this is like a simple @xmath60 minimization , but the points are not independent . we tried a fit to @xmath12 directly but found that it gave too much weight to the central regions and so instead we fit to @xmath234 $ ] so that the overall shape of the contours has an increased influence on the fit . the best - fit model parameters are listed in table[t : sp2 ] . the errors we quote are the rms spread in errors from fitting each mock catalogue in the same way . there are two key assumptions made in the construction of these models . firstly , although the contours match well at small scales , there are good reasons to believe that our linear theory model will not hold in the non - linear regime for @xmath235 . secondly , we have assumed the power - law model for @xmath7 and we have seen evidence that this is not completely realistic . using non - power law forms will also allow us to probe to larger scales . to test whether our result is robust to these assumptions we firstly reject the non - linear regime corresponding to @xmath236 . then , we use the shape of the hubble volume @xmath7 instead of a power - law , and finally we extend the maximum scale to @xmath237 . we showed in section [ s : inv ] that the hubble volume shape gives a good match to the data over the range @xmath238 ( the appendix gives the relevant equations for performing the @xmath15 infall calculation without a power - law assumption ) . we find that the best - fit parameters change very little with these changes but when using the hubble volume @xmath7 , the quality of the fit improves significantly . the best - fit model is compared to the data in fig . [ f : sps2 ] . notice the excellent agreement on small scales even though they are ignored in the fitting process . the best - fit parameters are listed in table [ t : sp2 ] , and we adopt these results as our final best estimates finding @xmath239 . if we repeat our analysis on the mock catalogues we find a mean value of @xmath240 ( cf . the expected value of @xmath186 , section [ s : mocks ] ) , showing that we can correctly determine @xmath15 using this type of fit . when fitting the mock catalogues it became clear that @xmath15 and @xmath18 are correlated in this fitting procedure , as we have seen already with other methods . we use the mock catalogues to measure the linear correlation coefficient , @xmath142 ( press 1992 ) , which quantifies this correlation , and find that , between @xmath15 and @xmath18 , @xmath241 . if we knew either parameter exactly , the error on the other would be smaller than quoted . we also tried other analytical forms for the correlation function and also different scale limits and found that some combinations shifted the results by @xmath242 . llll parameter & sgp & ngp & combined + + @xmath15 & @xmath243 & @xmath244 & @xmath245 + @xmath246 & @xmath247 & @xmath248 & @xmath249 + @xmath151 & @xmath250 & @xmath251 & @xmath252 + @xmath253 & @xmath254 & @xmath255 & @xmath256 + + @xmath15 & @xmath257 & @xmath258 & @xmath259 + @xmath246 & @xmath260 & @xmath261 & @xmath262 + @xmath151 & @xmath263 & @xmath264 & @xmath265 + @xmath253 & @xmath266 & @xmath267 & @xmath268 + + @xmath15 & @xmath269 & @xmath270 & @xmath271 + @xmath253 & @xmath272 & @xmath273 & @xmath274 + + @xmath15 & @xmath275 & @xmath276 & @xmath277 + @xmath253 & @xmath278 & @xmath279 & @xmath280 + we have now estimated the real - space clustering parameters using three different methods . in section [ s : xir_comp ] , we saw that the projection and inversion methods gave essentially identical results for @xmath122 and @xmath151 whereas using 2-d fits we get slightly higher values for @xmath122 . if @xmath7 was a perfect power - law the different methods would give unbiased results for the parameters , but we have seen evidence that this assumption is not true . the methods therefore , give different answers as a result of the different scales and weighting schemes used , as well as the vastly different treatments of the redshift - space distortions . it is worth contrasting our present results with those obtained in a previous 2dfgrs analysis ( peacock et al . this was based on the data available up to the end of 2000 : a total of 141402 redshifts . the chosen redshift limit was @xmath281 , yielding 127081 galaxies for the analysis of @xmath282 . the present analysis uses 165659 galaxies , but to a maximum redshift of 0.2 . because galaxies are given a redshift - dependent weight , this difference in redshift limit has a substantial effect on the volume sampled . for a given area of sky , changing the redshift limit from @xmath283 to @xmath281 changes the total number of galaxies by a factor of only 1.08 , whereas the total comoving volume within @xmath284 increases by a factor of 2 . allowing for the redshift - dependent weight used in practice , the difference in effective comoving volume for a given area of sky due to the variation in redshift limits becomes a factor of 1.6 . since the effective area covered by the present data is greater by a factor of @xmath285 , the total effective comoving volume probed in the current analysis is in fact 15% smaller than in the 2001 analysis ; this would suggest random errors on clustering statistics about 7% larger than previously . of course , the lower redshift limit has several important advantages : uncertainties in the selection function in the tail of the luminosity function are not an issue ( see norberg et al . 2002a ) ; also , the mean epoch of measurement is closer to @xmath286 . given that the sky coverage is now more uniform , and that the survey mask and selection function have been studied in greater detail , the present results should be much more robust . the other main difference between the present work and that of peacock ( 2001 ) lies in the method of analysis . the earlier work quantified the flattening of the contours of @xmath282 via the quadrupole - to - monopole ratio , @xmath287 . this is not to be confused with the quantity @xmath288 from section [ s : q ] , which uses an integrated clustering measure instead of @xmath289 . this is inevitably more noisy , as reflected in the error bar , @xmath290 , resulting from that method . the disadvantage of using @xmath287 directly , however , is that the ratio depends on the true shape of @xmath291 . in peacock 2001 ) , this was assumed to be known from the de - projection of angular clustering in the apm survey ( baugh & efstathiou 1993 ) ; in the present paper we have made a detailed internal estimate of @xmath291 , and considered the effect of uncertainties in this quantity . apart from this difference , the previous method of fitting to @xmath287 should , in principle , give results that are similar to our full fit to @xmath282 in section [ s : fits ] . the key issue in both cases is the treatment of the errors , which are estimated in a fully realistic fashion in the present paper using mock samples . the previous analysis used two simpler methods : an empirical error on @xmath287 was deduced from the ngp sgp difference , and correlated data were allowed for by estimating the true number of degrees of freedom from the value of @xmath60 for the best - fit model . this estimate was compared with a covariance matrix built from multiple realizations of @xmath282 using gaussian fields ; consistent errors were obtained . we applied the simple method of peacock ( 2001 ) to the current data , keeping the assumed apm @xmath291 , and obtained the marginalized result @xmath292 . the comparison with our best estimate of @xmath293 indicates that the systematic errors in the previous analysis ( from e.g. the assumed @xmath294 $ ] ) were not important , but that the previous error bars were optimistic by about 20% . there has been much discussion in the literature on whether or not the pairwise peculiar velocity dispersion , @xmath18 , is a function of projected separation , @xmath26 . many authors have used @xmath25-body simulations to make predictions for what might be observed . davis ( 1985 ) found that the pairwise velocity dispersion of cold dark matter remains approximately constant on small scales , decreases by about 20 - 30% on intermediate scales and is approximately constant again on large scales . cen , bahcall & gramann ( 1994 ) found a similar overall behaviour as did jenkins ( 1998 ) whose results are plotted in the left panel of fig . [ f : va ] as the solid line for a @xmath13 cosmology . the dashed line is from peacock & smith ( 2000 ) who used the halo model to predict the peculiar velocities for the galaxy distribution . kauffmann ( 1999 ) and benson ( 2000 ) used the gif simulations combined with semi - analytic models of galaxy formation and the galaxy predictions of benson ( 2000 ) are shown by the dotted line . these predictions generally assume @xmath295 , but there is evidence that @xmath296 could be @xmath297 lower than this ( spergel 2003 ) and so the pairwise velocity dispersions implied would also be lower . observationally , jing ( 1998 ) measured the pairwise velocity dispersion in the las campanas redshift survey and found no significant variation with scale . we note again that the errors for the lcrs ignore the effects of cosmic variance and are likely to be underestimates . zehavi ( 2002 ) used the sdss data and found that @xmath18 decreased with scale for @xmath298 . these observations are plotted in the right panel of fig . [ f : va ] . all these observations have assumed a functional form for the infall velocities ( or ` streaming ' ) and not used @xmath15 directly . we have already shown that proper consideration of the infall parameter is vital in such studies . indeed , zehavi ( 2002 ) say that their estimates of @xmath18 for @xmath299 depend significantly on their choice of streaming model . this factor , along with a dependence of @xmath18 on luminosity and galaxy type may help to explain the differences between the 2dfgrs and sdss results . the difference in results from section [ s : vfv ] which measured the value of @xmath18 at small @xmath26 ( @xmath300 ) , and from using the @xmath0 grid ( @xmath301 ) , which measures an average value , hints that there may be such an dependence of @xmath18 on @xmath26 in the 2dfgrs data . we test for variations in @xmath18 by repeating the fits described in section[s : fits ] using a global @xmath302 and @xmath151 but allowing @xmath18 to vary in each @xmath26 slice . the results are shown in fig.[f : va ] , compared with the results from other surveys , and numerical simulations as discussed above . the value of @xmath301 obtained from the 2-d fit for scales @xmath303 is close to the value at @xmath304 where most of the signal is coming from . the value of @xmath300 obtained from the fourier transform technique agrees well with the results found for @xmath305 . the values of @xmath306 and @xmath151 are essentially unchanged when fitting in this way . we note again that the effects of the infall must be properly taken into account in these measurements . we also note that we used our linear , power - law model on all scales , but we have seen that this is a reasonable approximation on non - linear scales . we see that the overall shape of the 2dfgrs results are fairly consistent with , though slightly flatter than the semi - analytic predictions , but the amplitude is certainly a little different , which could be due to the value of @xmath296 used in the models , as discussed above . we also plot the mean of the mock catalogue results ( solid line ) , and the results of a simulated catalogue ( dashed line ) of yang ( 2003 , with @xmath307 ) and these match the real data well . we take the value of @xmath15 measured from the multi - parameter best fit to @xmath0 , @xmath309 which is measured at the effective luminosity , @xmath310 , and redshift , @xmath311 , of our survey sample . in section [ s:2dfdata ] we quoted these values , which are the applicable mean values when using the @xmath77 weighting and redshift cuts employed , as @xmath312 and @xmath2 . we also note here that if we adopt an @xmath313 geometry we find that @xmath314 , within the quoted 1@xmath26 errors . the redshift distortion parameter can be written as , @xmath315 where @xmath316 , and @xmath317 is the linear fluctuation growth factor and @xmath18 is the expansion factor . a good approximation for @xmath318 , at all @xmath319 , in a flat universe , was given by lahav ( 1991 ) , @xmath320 and so to constrain @xmath29 from these results we need an estimate of @xmath30 . there have been two recent papers describing such measurements . verde ( 2002 ) measured @xmath321 from an analysis of the bispectrum of 2dfgrs galaxies . their results depend strongly on the pairwise peculiar velocity dispersion , @xmath18 , assumed in their analysis . they used the result of peacock ( 2001 ) , who found @xmath322 , somewhat lower than our new value of @xmath323 . to derive @xmath29 using these results would not therefore be consistent and so a new bispectrum analysis is in preparation . lahav ( 2002 ) combined the estimate of the 2df power spectrum , @xmath324 ( percival 2001 ) , with pre-_wmap _ results from the cmb to obtain an estimate of @xmath30 , but this value is also dependent on @xmath29 . their likelihood contours , the optical depth of reionisation , as @xmath325 . the plotted results do not include this effect , which could be significant , and this is discussed further in section [ s : omcomp ] . ] are reproduced in fig.[f : omvb ] , as the dashed lines . they also introduced a ` constant galaxy clustering ' model for the evolution of @xmath30 with @xmath319 . following these equations we can evolve our measured @xmath15 to the present day and estimate @xmath326 and these contours are shown by the solid lines in fig.[f : omvb ] . these are in good agreement with , and orthogonal to , those of lahav ( 2002 ) . we note that the above analysis is independent of luminosity as we examine everything at the effective luminosity of the survey , @xmath310 . from the correlation functions in different volume - limited samples of 2dfgrs galaxies , norberg ( 2001 ) found a luminosity dependence of clustering of the form ( cf . [ e : r0star ] ) , @xmath327 which gives an estimate for the bias of the survey galaxies , @xmath328 ( using @xmath166 ) , where @xmath329 is the bias of @xmath330 galaxies . if this bias relation holds on the scales considered in this paper then @xmath15 will be increased by the same factor of 1.06 , @xmath331 and evolving @xmath15 in a ` constant galaxy clustering ' model ( lahav 2002 ) then , @xmath332 which we choose as a fiducial point to allow comparisons with other surveys with different effective luminosities and redshifts . lahav ( 2002 ) obtained , @xmath333 , in their combined 2dfgrs and cmb analysis , completely consistent with the our result . percival ( 2002 ) combined the 2dfgrs power spectrum with the pre-_wmap _ cmb data , assuming a flat cosmology and found @xmath334 . these measurements of @xmath29 are also consistent with a different estimation from the 2dfgrs and cmb ( efstathiou 2002 ) and from combining the 2dfgrs with cosmic shear measurements ( brown 2003 ) . also plotted in figure [ f : omvb ] is the recent result from the analysis of the _ wmap _ satellite data . spergel ( 2003 ) found @xmath335 using _ wmap _ data alone , although there are degeneracies with other parameters . it is clear that this is completely consistent with the other plotted contours . spergel ( 2003 ) also found that the epoch of reionisation , @xmath336 , which would reduce the value of @xmath30 found by lahav ( 2002 ) by about @xmath337 , still in good agreement with the results in this paper . in this paper we have measured the correlation function , and various related quantities using 2dfgrs galaxies . our main results are summarised as follows : 1 . the spherical average of @xmath0 gives the redshift - space correlation function , @xmath4 , from which we measure the redshift space clustering length , @xmath5 . at large and small scales @xmath4 drops below a power law as expected , for instance , in the @xmath13 model . 2 . the projection of @xmath0 along the @xmath27 axis gives an estimate of the real - space correlation function , @xmath7 , which on scales @xmath338 can be fit by a power - law @xmath8 with @xmath339 , @xmath340 . at large scales , @xmath7 drops below a power - law as expected , for instance , in the @xmath13 model . the ratio of real and redshift - space correlation functions on scales of @xmath14 reflects systematic infall velocities and leads to an estimate of @xmath188 . the quadrupole moment of @xmath0 on large scales gives @xmath195 . 4 . comparing the projections of @xmath0 along the @xmath27 and @xmath26 axes gives an estimate of the distribution of random pairwise peculiar velocities , @xmath17 . we find that large - scale infall velocities affect the measurement of the distribution significantly and can not be neglected . using @xmath341 , we find that @xmath17 is well fit by an exponential with pairwise velocity dispersion , @xmath342 , at small @xmath26 . a multi - parameter fit to @xmath0 simultaneously constrains the shape and amplitude of @xmath7 and both the velocity distortion effects parameterized by @xmath15 and @xmath18 . we find @xmath239 and @xmath20 , using the hubble volume @xmath7 as input to the model . these results apply to galaxies with effective luminosity , @xmath343 and at an effective redshift , @xmath344 . we also find that the best fit values of @xmath15 and @xmath18 are strongly correlated . we evolve our value for the infall parameter to the present day and critical luminosity and find @xmath345 . our derived constraints on @xmath29 and @xmath30 are consistent with a range of other recent analyses . our results show that the clustering of 2dfgrs galaxies as a whole is well matched by a low density @xmath13 simulation with a non - linear local bias scheme based on the smoothed dark - matter density field . nevertheless , there are features of the galaxy distribution which require more sophisticated models , for example the distribution of pairwise velocities and the dependence of galaxy clustering on luminosity or spectral type . the methods presented have also been used on sub - samples of the 2dfgrs , split by their spectral type ( madgwick 2003 ) . the 2df galaxy redshift survey has been made possible through the dedicated efforts of the staff of the anglo - australian observatory , both in creating the 2df instrument and in supporting it on the telescope . we thank nelson padilla , andrew benson , sarah bridle , yipeng jing , xiaohu yang and robert smith for providing their results in electronic form . we also thank the referee for valuable comments . baugh , c.m . , efstathiou , g. , 1993 , mnras , 265 , 145 benson , a.j . , baugh , c.m . , cole , s. , frenk , c.s . , lacey , c.g . , 2000 , mnras , 316 , 107 brown , m.l . , taylor , a.n . , bacon , d.j . , gray , m.e . , dye , s. , meisenheimer , k. , wolf , c. , 2003 , mnras , 341 , 100 cen , r. , bahcall , n.a . , gramann , m. , 1994 , apj , 437 , 51 cole , s. , hatton , s. , weinberg , d.h . , frenk , c.s . , 1998 , mnras , 300 , 945 colless , m. & the 2dfgrs team , 2001 , mnras , 328 , 1039 davis , m. , peebles , p.j.e . , 1983 , apj , 208 , 13 davis , m. , efstathiou , g. , frenk , c.s . , white , s.d.m . , 1985 , apj , 292 , 371 diaferio , a. , geller , m. , 1996 , apj , 467 , 19 efstathiou , g. , 1988 , in proc . 3rd _ iras _ conf . , comets to cosmology , ed . a. lawrence ( new york , springer ) , p. 312 efstathiou , g. & the 2dfgrs team , 2002 , mnras , 330 , 29 evrard a.e . , 2002 , apj , 573 , 7 guzzo , l. , 2000 , a&a , 355 , 1 hamilton , a.j.s . , 1992 , apj , 385 , l5 hamilton , a.j.s . , 1993 , apj , 417 , 19 hawkins , e. , maddox , s. , branchini , e. , saunders , w. , 2001 , mnras , 325 , 589 jenkins , a. , 1998 , apj , 499 , 20 jing , y.p . , mo , h.j . , brner , g. , 1998 , aj , 494 , 1 kaiser , n. , 1987 , mnras , 227 , 1 kauffmann , g. , colberg , j.m . , diaferio , a. , white , s.d.m . , 1999 , mnras , 303 , 188 . lahav , o. , lilje , p.b . , primack , j.r . , rees , m.j . , 1991 , mnras , 251 , 128 lahav , o. & the 2dfgrs team , 2002 , mnras , 333 , 961 landy , s.d . , szalay , a.s . , 1993 , apj , 388 , 310 landy , s.d . , szalay , a.s . , broadhurst , t. , 1998 , apj , 494 , l133 ( lsb98 ) landy , s.d . , 2002 , apj , 567 , l1 ( l02 ) lewis , i.j . , 2002 , mnras , 333 , 279 limber , d.n . , 1954 , apj , 119 , 655 lin , h. , kirshner , r.p . , shectman , s.a . , landy , s.d . , oemler , a. , tucker , d.l . , schechter , p.l . , 1996 , apj , 464 , 60 loveday , j. , efstathiou , g. , peterson , b.a . , maddox , s.j . , 1992 , apj , 400 , 43 loveday , j. , maddox , s.j . , efstathiou , g. , peterson , b.a . , 1995 , apj , 442 , 457 maddox , s.j . , efstathiou , g. , sutherland , w.j . , 1990 , mnras , 246 , 433 maddox , s.j . , efstathiou , g. , sutherland , w.j . , 1996 , mnras , 283 , 1227 madgwick , d.s . & the 2dfgrs team , 2003 , mnras , submitted , astro - ph/0303668 magliochetti , m. , porciani , c. , 2003 , mnras , submitted , astro - ph/0304003 marzke , r.o . , geller , m.j . , da costa , l.n . , huchra , j.p . , 1995 , aj , 110 , 477 norberg , p. & the 2dfgrs team , 2001 , mnras , 328 , 64 norberg , p. & the 2dfgrs team , 2002a , mnras 336 , 907 norberg , p. & the 2dfgrs team , 2002b , mnras , 332 , 827 padilla , n. , baugh , c.m . , 2003 , mnras , submitted , astro - ph/0301083 peacock , j.a . , smith , r.e . , 2000 , mnras , 318 , 1144 peacock , j.a . & the 2dfgrs team , 2001 , nature , 410 , 169 peebles , p.j.e . , 1980 , _ the large scale structure of the universe _ , princeton univ . press , princeton , nj percival , w. & the 2dfgrs team , 2001 , mnras , 327 , 1297 percival , w. & the 2dfgrs team , 2002 , mnras , 337 , 1068 press , w. , teukolsky , s. , vetterling , w. , flannery , b. , 1992 , _ numerical recipes _ , cambridge university press , cambridge ratcliffe , a. , shanks , t. , parker , q.a . , fong , r. , 1998 , mnras , 296 , 191 saunders , w. , rowan - robinson , m. , lawrence , a. , 1992 , mnras , 258 , 134 ( s92 ) sheth , r.k . , 1996 , mnras , 279 , 1310 spergel , d.n . ( the _ wmap _ team ) , 2003 , apj , submitted , astro - ph/0302209 tucker , d.l . , 1997 , mnras , 285 , 5 verde , l. & the 2dfgrs team , 2002 , mnras , 335 , 432 yang , x. , mo , h.j . , jing , y.p . , van den bosch , f. , chu , y , 2003 , mnras , submitted , astro - ph/0303524 zehavi , i. ( the sdss collaboration ) , 2002 , apj , 571 , 172 zehavi , i. ( the sdss collaboration ) , 2003 , apj , submitted , astro - ph/0301280 kaiser ( 1987 ) pointed out that the coherent infall velocities take a simple form in fourier space , @xmath346 hamilton ( 1992 ) completed the translation of these results into real space , @xmath347 ^ 2 \xir,\ ] ] which reduces to @xmath348 where in general , @xmath349 @xmath350,\ ] ] @xmath351,\ ] ] and @xmath352 @xmath353 in the case of a power law form for @xmath7 these equations reduce to the form shown in eqns . [ e : xi0 ] to [ e : xi4 ] . in the case of non - power law forms for the real - space correlation function these integrals must be performed numerically .	we present a detailed analysis of the two - point correlation function , @xmath0 , from the 2df galaxy redshift survey ( 2dfgrs ) . the large size of the catalogue , which contains @xmath1 redshifts , allows us to make high precision measurements of various properties of the galaxy clustering pattern . the effective redshift at which our estimates are made is @xmath2 , and similarly the effective luminosity , @xmath3 . we estimate the redshift - space correlation function , @xmath4 , from which we measure the redshift - space clustering length , @xmath5 . we also estimate the projected correlation function , @xmath6 , and the real - space correlation function , @xmath7 , which can be fit by a power - law @xmath8 , with @xmath9 , @xmath10 . for @xmath11 , @xmath12 drops below a power - law as , for instance , is expected in the popular @xmath13 model . the ratio of amplitudes of the real and redshift - space correlation functions on scales of @xmath14 gives an estimate of the redshift - space distortion parameter @xmath15 . the quadrupole moment of @xmath0 on scales @xmath16 provides another estimate of @xmath15 . we also estimate the distribution function of pairwise peculiar velocities , @xmath17 , including rigorously the significant effect due to the infall velocities , and find that the distribution is well fit by an exponential form . the accuracy of our @xmath0 measurement is sufficient to constrain a model , which simultaneously fits the shape and amplitude of @xmath7 and the two redshift - space distortion effects parameterized by @xmath15 and velocity dispersion , @xmath18 . we find @xmath19 and @xmath20 , though the best fit values are strongly correlated . we measure the variation of the peculiar velocity dispersion with projected separation , @xmath21 , and find that the shape is consistent with models and simulations . this is the first time that @xmath15 and @xmath17 have been estimated from a self - consistent model of galaxy velocities . using the constraints on bias from recent estimates , and taking account of redshift evolution , we conclude that @xmath22 , and that the present day matter density of the universe , @xmath23 , consistent with other 2dfgrs estimates and independent analyses . [ firstpage ] galaxies : statistics , distances and redshifts - large scale structure of universe - cosmological parameters - surveys
the collision of relativistic heavy ions remains the only practical way to study hot and dense strongly interacting matter in the laboratory , and a vigorous experimental program at the energy frontier is currently being pursued at rhic , and another will soon begin at the lhc . the study of in - medium properties of vector mesons is a topic of continuing interest , as their spectral properties can be linked directly to the electromagnetic emissivity of the strongly interacting medium @xcite . as real and virtual photons leave their production site essentially unscathed , they represent penetrating probes that can reveal what conditions existed when they were emitted . note however that electromagnetic signals are continuously being generated : a reliable interpretation of experimental measurements still hinges on our understanding of the space - time evolution of the source . many recent theoretical studies have concentrated on the lower mass vector mesons i.e. @xmath1 and @xmath2 ( see @xcite and references therein ) , owing partly to the availability of precise heavy ion data @xcite in this invariant mass range . in this work , we calculate the in - medium modifications of the properties of the @xmath0 meson . this resonance offers the advantage of a well - defined and narrow distribution , making it a good candidate for precision measurements . on the theory front , some previous investigations have been pursued in various limits of temperature and density with different techniques . a non - exhaustive list of those tools includes : qcd sum rules @xcite , effective hadronic lagrangians @xcite , and the nambu - jona - lasinio lagrangian @xcite . estimates of the kinematical broadening of the @xmath0 have also been done @xcite . experimentally , rhic measurements have concentrated up to now on the bulk behavior of the @xmath0 @xcite , even though some spectral information is emerging @xcite . in this context , a tantalizing result is that of the kek - ps e325 collaboration @xcite which claims to have measured the first experimental evidence of in - medium modification of the @xmath0 meson in nuclear matter . the goal of this work is not yet to directly compare with experimental measurements of the properties of the @xmath0 in relativistic nuclear collisions . as a first step , the size of expected in - medium modifications should be investigated at finite temperatures and densities , in a simpler equilibrium setting : we report here on such an investigation . the paper is organized in the following way : in section ii we discuss the formalism used to calculate the self - energy , together with the connection to the forward scattering amplitude ( fsa ) . in section iii , the methodology to calculate the fsa is given , and the next section features results . we conclude in section v. for a @xmath0 meson interacting with hadron @xmath3 in the medium , the retarded self - energy can be expressed as @xcite @xmath4 f_{\phi a}\left(\frac{m_\phi}{m_a}\omega\right)\ ] ] to leading order in the density of scatterers , in the rest frame of @xmath3 , and on the mass shell of the @xmath0 meson . in the above , @xmath5 is the momentum of the @xmath0 meson , @xmath6 , @xmath7 is the amplitude for the forward scattering of @xmath0 with the field of type @xmath3 , and @xmath8 is the temperature . also , @xmath9 , with @xmath3 a boson . if @xmath3 is a fermion with a chemical potential @xmath10 , the argument of the logarithm becomes @xmath11 @xcite . turning first to the zero temperature part of the self - energy , we identify two contributions . the first is calculated through the interaction of the @xmath0 meson with a @xmath1 and @xmath12 , and is illustrated in fig . [ pic : phi_self ] ( a ) . meson . ] this interaction is well described by the wess - zumino lagrangian term , with @xmath2-@xmath0 mixing @xcite : @xmath13 where @xmath14 using the wess - zumino interaction , we calculate the vacuum self - energy of the @xmath0 meson : @xmath15 + c \label{eq : self - rho - pi}\end{aligned}\ ] ] using the experimentally measured branching ratio , the coupling constant in a given isospin channel is determined to be @xmath16 mev , while the mixing angle is @xmath17 @xcite . note that @xmath18 , @xmath19 is the invariant mass of the @xmath0 meson , and @xmath20 is the euler - mascheroni constant . the renormalization constant @xmath21 is chosen such that @xmath22=0 $ ] . the second contribution to the vacuum self - energy is given in fig . [ pic : phi_self ] ( b ) . such an interaction is described by the lagrangian of the form @xcite : @xmath23 where @xmath24 in the complex charged kaon field , @xmath25 is the @xmath0 field strength and @xmath26 is the covariant derivative . the vacuum self - energy for the @xmath27 interaction is : @xmath28 + \frac{8 m^2_k}{m^2 } + \bar{c}\right\ } \label{eq : self - kk}\ ] ] where the kaon mass is either @xmath29 gev or @xmath30 gev for the charged and neutral kaons respectively . the values of the coupling are @xmath31 and 1.682 for the charged and neutral kaons @xcite . finally , @xmath32 is determined the same way as @xmath21 in eq . ( [ eq : self - rho - pi ] ) . the net self - energy is given by summing over all target species and including the vacuum contributions : @xmath33 the individual contributions to the self - energy contain the appropriate spin / isospin symmetry factors . note that the functional dependence of the self - energy in the vacuum case is different than that at finite temperature : the vacuum part of @xmath34 can solely depend on the invariant mass , @xmath35 . the matter parts do not have such a restriction and in general depend on both @xmath36 and @xmath37 . however , in this study , the scattering amplitudes are evaluated on the mass shell of the @xmath0 meson . therefore , the vacuum contribution is evaluated at @xmath38 and the matter self - energies are only @xmath37-dependent . we only consider interactions with pions ( @xmath12 ) and nucleons ( @xmath39 ) as scatterings of the @xmath0 meson contributing to its self - energy . to describe @xmath0 s self - energy , a two - component approach @xcite is used ( see also @xcite ) : while ordinary reggeons are dual to s - channel resonances , the pomeron is dual to the background under the resonances . the low energy fsa is therefore composed of two parts , namely a resonance contribution and a background pomeron term . in the center of mass ( cm ) frame the low energy fsa reads @xcite @xmath40 here the sum ranges over resonances that decay into the @xmath0 meson and the particle @xmath3 , which is either a nucleon or a pion . the mass of the resonance @xmath41 is @xmath42 and its total width is @xmath43 . as usual , @xmath44 is the mandelstam variable , and the magnitude of the center of mass momentum can be re - expressed in terms of masses and @xmath44 . spin / isospin statistics are taken care of by using the averaging factor @xmath45 , with @xmath46 being the spin of particle @xmath47 , and @xmath48 , the isospin . this averaging procedure will wash out the distinction between longitudinal and transverse spin directions , and also the difference between different charge states . the partial width for a resonance to decay into the @xmath49 channel is denoted by @xmath50 in eq.([eq : f_low ] ) . in the case were @xmath3 is a nucleon , the effective width can not be obtained directly from experiment @xcite , as no resonances that decay directly into @xmath0 and @xmath39 have yet been observed @xcite . a scaling law first proposed by lipkin @xcite in order to infer the effective width of @xmath50 is used . lipkin found that the ratio @xmath51 is a constant that quantifies the experimental deviation from the ideal su(3 ) octet - singlet mixing . we will be using the average value of @xmath52 ( averaged over the range @xmath53 gev ) , henceforth denoted @xmath54 . _ @xcite have recently obtained @xmath55 via analysis of @xmath0 and @xmath2 production in @xmath56 and @xmath57 reactions . it is assumed that resonances that can connect to an @xmath58 state @xcite will also have a @xmath59 interaction , with the appropriate kinematic adjustments . however , those resonances will have a reduced branching ratio @xmath60 by @xmath54 , i.e. @xmath61 . similarly to what was previously done @xcite , @xmath62 is expressed as : @xmath63 where @xmath64 is the total width of the resonance @xmath41 , @xmath65 is the branching ratio of the decay @xmath66 , we take the average value @xmath67 , @xmath68 \left [ m^{2}_{r } - \left ( m_\phi - m_a \right)^2 \right ] } } { m_{r } } $ ] , and @xmath69 is the smallest relative angular momentum between @xmath0 and @xmath39 . reactions proceeding via the strong interaction conserve parity ; we use this fact to determine @xmath69 . the parity conservation can be stated as @xmath70 where @xmath71 is the parity of the resonance , @xmath72 is the parity of the @xmath0 meson , @xmath73 is the parity of the nucleon , and @xmath74 is the parity of the wavefunction describing the relative angular momentum of @xmath0 and @xmath39 . parity conservation determines only whether @xmath69 is even or odd . to determine minimum @xmath69 , we need to use spin and angular momentum addition rule . this rule looks as follows : @xmath75 where @xmath76 is the total spin operator of the resonance @xmath41 , @xmath77 is the angular momentum operator , @xmath78 is the spin operator of the @xmath0 meson , and @xmath79 is the spin of the nucleon . the non - vanishing clebsch - gordan coefficients will determine the possible values @xmath69 can take in the even / odd number region delimited by parity conservation equation . we keep only the smallest @xmath69 that satisfies both spin and parity conditions . the above approach to determine spin and parity is general and also applicable for interactions with pions ( instead of nucleons ) . besides the above - threshold resonances , it is also necessary to include subthreshold resonances for they also make a contribution to the fsa . their contribution however is relatively small . indeed , including subthreshold resonances changes the imaginary part of the fsa by at most 2% in the @xmath80 gev energy range . the change to the real part of the fsa due to subthreshold effects is significantly larger reaching 50% in the same energy range . hence it is important to include the subthreshold resonances in the fsa . note that the overall magnitude of the real part of the fsa in that region is quite small ( see fig . [ pic : re_im_f ] ) , so the aforementioned 50% change is not as dramatic as might first appear . to estimate their widths we assume that the vector meson dominance model ( vmd ) is valid . as in @xcite , we write the partial width of the subthreshold resonances as @xmath81 and @xmath82 , where @xmath83 is the @xmath84 center of mass momentum . vmd allows us to relate @xmath85 ( with @xmath86 ) to @xmath87 as follows @xcite : @xmath88 one may evaluate the value of the couplings @xmath89 by first writing down the vmd lagrangian @xcite which governs the electromagnetic decay of vector mesons : @xmath90 where @xmath91 . the fermionic field for lepton @xmath69 is @xmath92 , @xmath93 is the photon field with the corresponding field strength tensor @xmath94 , @xmath95 are the different vector meson fields , namely @xmath1 , @xmath2 and @xmath0 fields , with the field strength tensor written in an analogous fashion to the photon case and @xmath96 is the electromagnetic fine - structure constant . finally , the different @xmath97 ( with @xmath47 generic ) are the masses of the different particles involved . the coupling constants are then easily determined by evaluating the decay width into a lepton pair @xmath98 which stems from the feynman amplitude of fig . [ pic : em_decay ] . the values obtained are @xmath99 , @xmath100 , and @xmath101 @xcite . inspired by the quark model it is estimated that @xmath102 @xcite . further , using the scaling law proposed by lipkin , the relation @xmath103 is invoked . these statements allow to simplify eq . ( [ subth_gamma ] ) : @xmath104 where @xmath105 , with @xmath106 being the experimental values quoted in ref . a summary of all the resonances that decay into @xmath59 is presented in table [ table : phi_n ] . .baryon resonances , both above and below threshold , included in the calculation of the @xmath59 fsa . [ cols="^,^,^,^ " , ] [ table : width_phi_t ] we have calculated the mass shift and width broadening as a function of momentum for the @xmath0 meson using the scheme developed in @xcite . this approach has the great advantage to be model - independent since experimental data are taken as input to calculate the fsa . our results show that pion interactions contribute to reduce the pole mass of the @xmath0 whereas nucleons act in the opposite way hence increasing the pole mass . the mass shift overall is small and about a few tens of mev at most . also , we have found a significant increase of the width of the @xmath0 meson from its vacuum value of 4.26 mev . hence the spectral density , which is related to the dilepton production , was considerably broadened . these results will be important for understanding possible in - medium modifications of the @xmath0 meson in strongly interacting environments . it is a pleasure to thank j. i. kapusta , p. lichard , and a.t . martell for usefull suggestions and discussions . this work was supported in part by the natural sciences and engineering research council of canada , and in part by the fonds nature et technologies of quebec . 00 joseph i. kapusta and charles gale , _ finite - temperature field theory : principles and applications _ ( cambridge university press , cambridge , uk , 2006 ) . c. gale and j. i. kapusta , nucl . b357 , 65 ( 1991 ) . r. rapp and j. wambach , adv . phys . * 25 * , 1 ( 2000 ) . v. l. eletsky , m. belkacem , p. j. ellis , and j. i. kapusta , phys . c * 64 * , 035202 ( 2001 ) . a. t. martell and p. j. ellis , phys . c * 69 * , 065206 ( 2004 ) . j. ruppert , c. gale , t. renk , p. lichard and j. i. kapusta , phys . lett . * 100 * , 162301 ( 2008 ) . h. van hees and r. rapp , arxiv:0711.3444 [ hep - ph ] . d. miskowiec [ ceres collaboration ] , nucl . a * 774 * , 43 ( 2006 ) . r. arnaldi _ et al . _ [ na60 collaboration ] , phys . * 96 * , 162302 ( 2006 ) . t. hatsuda and s. h. lee , phys . c * 46 * , 34 ( 1992 ) . k. l. haglin and c. gale , nucl . b * 421 * , 613 ( 1994 ) . f. klingl , t. waas and w. weise , phys . b * 431 * , 254 ( 1998 ) . e. oset and a. ramos , nucl . phys . a * 679 * , 616 ( 2001 ) . d. cabrera and m. j. vincente vacas , phys . c * 67 * , 045203 ( 2003 ) . j. p. blaizot and r. mendez galain , phys . b * 271 * , 32 ( 1991 ) . p. z. bi and j. rafelski , phys . b * 262 * , 485 ( 1991 ) . l. holt and k. haglin , j. phys . g : nucl . part . phys . * 31 * 245 ( 2005 ) . a. sibirtsev , h .- w . hammer , u .- g . meissner and a.w . thomas , nucl - th/0606044 . blyth , j. phys . g * 32 * , s461 ( 2006 ) ; y. g. ma , j. phys . g * 32 * , s373 ( 2006 ) . s. afanasiev _ et al . _ [ phenix collaboration ] , arxiv:0706.3034 [ nucl - ex ] . r. muto et al . , phys . lett . * 98 * , 042501 ( 2007 ) . s. jeon and p. j. ellis , phys . d * 58 * , 045013 ( 1998 ) . v. l. eletsky , b. l. ioffe , and j. i. kapusta , eur . j. a * 3 * , 381 ( 1998 ) ; v. l. eletsky and j. i. kapusta , phys . rev . c * 59 * , 2757 ( 1999 ) . c. gale , d. seibert and j. kapusta , phys . d * 56 * , 508 ( 1997 ) . k. haglin , and c. gale , nucl . phys b * 421 * , 613 ( 1994 ) . h. harari , phys . . lett . * 20 * , 1395 ( 1968 ) . d. b. collins , _ an introduction to regge theory and high energy physics _ ( cambridge university press , cambridge , uk , 1977 ) . w .- m.yao _ ( particle data group ) , j. phys . g * 33 * , 1 ( 2006 ) . actually , there is only one resonance has been observed to decays into a @xmath0 and a nucleon , the @xmath107 . however , its partial width in this channel is small and has been neglected . h. j. lipkin , phys . b * 60 * , 371 ( 1976 ) . a. sibirtsev , j. haidenbauer , and u .- g . meissner , eur . j. a * 27 * , 263 ( 2006 ) ; a. sibirstev and w. cassing eur . j. a * 7 * , 407 ( 2000 ) . ioffe , v.a . khoze , and l.n . lipatov , _ hard processes volume 1 : phenomenology of quark - parton model _ ( north - holland , amsterdam , 1984 ) . v. shklyar , g. penner and u. mosel , eur . j. a * 21 * 445 ( 2004 ) . a. donnachie , p.v . landshoff , phys . b * 296 * 227 ( 1992 ) . g. j. gounaris and j. j. sakurai , phys . 21 , 244 ( 1968 ) . l. d. mclerran and t. toimela , phys . rev . * d 31 * , 545 ( 1985 ) . h. a. weldon , phys . d * 42 * , 2384 ( 1990 ) . c , * 63 * 054907 ( 2001 ) .	the spectral density of the @xmath0 meson in a hot bath of nucleons and pions is calculated by relating the vector meson self - energy to the forward scattering amplitude ( fsa ) , which is constrained by experimental data . dispersion techniques are used to verify the relationship between real and the imaginary parts of the in - medium self energy . the position of the spectral peak of the @xmath0 meson is found to be shifted from its vacuum position by only a small amount , but its width is considerably increased .
since the launch of the hubble space telescope ( hst ) many imaging studies of the narrow line regions ( nlr ) of active galactic nuclei ( agn ) have been carried out . these studies have shown that the emission line gas often has a complex morphology , frequently taking the form of a bicone centered on the galaxy nucleus . ( _ e.g. _ ngc 4151 , evans et al . , 1993 , boksenberg _ et al . _ 1995 , ngc 1068 , evans _ et al . _ 1991 , see also the archival study by schmitt & kinney 1996 ) . in the standard model for an agn , a dense molecular torus with a radius of a few parsecs surrounds the nucleus and collimates the radiation field ( _ e.g. _ antonucci 1993 ) . according to the model , differences in the continuum and emission line spectra , which form the basis for classification of seyferts and other types of agn , can be explained largely by differences in the orientation of the torus to our line of sight . for example , in type 1 seyfert galaxies our viewing angle is close to the symmetry axis of the torus allowing a direct view of the broad line region ( blr ) and the nuclear continuum source , while in type 2 seyfert galaxies our vantage point lies closer to the plane of the torus which then blocks a direct view of the inner regions . in many instances the nlr morphology and kinematics appear closely linked to the radio structure , particularly in seyfert galaxies with linear or jet - like radio sources . in these objects the line emitting gas is often found to be cospatial with the radio jets and there is also kinematic evidence for physical interaction between the jets and the nlr gas ( capetti _ et al . _ 1999 , whittle _ et al . _ , 1988 ) . the suggestion has been made that expansion of the radio plasma into the host galaxy s interstellar medium produces fast shock waves which emit a hard continuum and ultimately provide the dominant source of ionizing photons ( taylor , dyson , & axon , 1992 , sutherland , bicknell & dopita , 1993 ) . the degree to which photoionization by a nuclear continuum source or by autoionizing shocks contributes to the overall energetics of the nlr has been the subject of some debate . in principal one can distinguish between them spectroscopically by studying the spatially resolved kinematics and the physical conditions of the gas as revealed by the relative intensities of specific emission lines . the space telescope imaging spectrograph ( stis ) is ideally suited for this type of study . we have therefore undertaken a detailed investigation of the kinematics and physical conditions across the nlr of ngc 4151 , one of the nearest seyfert galaxies . evidence for outflow and photo - ionization cones in the nlr of ngc 4151 was presented by schulz ( 1988 , 1990 ) based on ground - based longslit spectroscopy . peculiar flat - topped and double - peaked emission line profiles were observed to the se and nw between 2@xmath4 and 6@xmath4 from the nucleus and are most consistent with outflow models . schulz ( 1990 ) suggests that the outflow is driven either by a wind related to the active nucleus or by an expanding radio plasmon . the nlr kinematics in ngc 4151 have been studied in detail using slitless spectroscopy from stis ( hutchings _ et al . _ , 1998 , kaiser _ et al . _ , 1999 , and hutchings _ et al . _ , 1999 ) . these observations reveal three distinct kinematic components : one consisting of low velocity clouds ( @xmath5 km s@xmath0 ) , primarily in the outer nlr following the rotation of the host galaxy disk , a second consisting of moderately high velocity clouds ( @xmath6 km s@xmath0 ) most likely associated with radial outflow within the biconical morphology and a third component of fainter but much higher velocity clouds ( @xmath7 km s@xmath0 ) which is also outflowing but not restricted to the biconical flow of the intermediate velocity component . no evidence for higher velocities in the vicinity of the radio knots was found suggesting that the radio jet has minimal influence on the nlr kinematics . a somewhat different conclusion was drawn by winge _ ( 1999 ) primarily using longslit spectroscopy with hst s faint object camera . they claim evidence for strong interaction between the radio jet and the nlr gas . furthermore , after subtracting the influence of the radio jet and galaxy rotation on the kinematics , they suggest that the residual motion is the rotation of a thin disk of gas on nearly keplerian orbits beyond [email protected] ( 60 pc using their linear scale ) around an enclosed mass of @xmath9 . interior to 60 pc the velocities turn over suggesting that the mass is extended , and , if their interpretation is correct , they are able to place upper limits on the mass of a nuclear black hole of @xmath10 . in this paper we present the initial results from our low resolution , longslit spectroscopy . a second paper presents a detailed photoionization model using the emission line ratios presented here ( kraemer _ et al . _ 1999 , paper ii ) . section [ obs ] presents the observations and describes the data reduction procedures including correction for scattered light from the seyfert nucleus . section [ ana ] describes the results of the kinematic and preliminary line ratio analyses . in section [ disc ] we discuss the results in terms of different nlr models . we summarize our results and conclusions in section [ conc ] . longslit spectroscopy of ngc 4151 was obtained with stis on board hst . four low dispersion gratings , g140l , g230 lb , g430l and g750l , were used producing spectra ranging from the uv at 1150 to the near - infrared at 10,270 . note that the g230 lb mode , which uses the ccd detector , was used instead of the g230l , due to the bright object protection limits imposed on use of the mama detectors . two slit alignments were chosen to cover regions of specific interest and as many of the bright emission line clouds as possible . the first position was chosen to pass through the nucleus at position angle 221@xmath11 , while the second was offset from the nucleus by [email protected] to the south at position angle 70@xmath11 . figure [ slits ] shows the slit apertures drawn on the wfpc-2 narrow band image of the [ oiii ] @xmath25007 emission line structure obtained from the hst archives ( proposal i d 5124 , principal investigator h. ford ) . the [email protected] slit was used to preserve spectral resolution , given here for each of the four gratings assuming an extended source ( the emission line clouds are generally resolved along the slit ) : 2.4 for g140l , 2.7 for g230 lb , 5.5 for g430l , and 9.8 for g750l ( woodgate _ et al . _ 1998 , kimble _ et al . _ 1998 , baum _ et al . _ 1998 ) . a log of the observations is presented in table 1 . one set of observations failed and as a result no g140l spectrum was available for p.a . 70@xmath11 . the spectra were reduced using the idl software developed at nasa s goddard space flight center for the instrument definition team ( lindler _ et al . cosmic ray hits were identified and removed from observations using the ccd detector ( g230 lb , g430l , and g750l ) by combining the multiple images obtained at each visit in each spectroscopic mode . hot or warm pixels ( identified in stis dark images ) were replaced by interpolation in the dispersion direction . wavelength calibration exposures obtained after each science observation were used to correct the wavelength scale for zero - point shifts . the spectra were also geometrically rectified and flux - calibrated to produce a constant wavelength along each column ( the spatial direction ) and fluxes in units of ergs s@xmath0 @xmath12 @xmath0 per cross - dispersion pixel . spectra obtained at the same position angle and spectroscopic mode were combined to increase the signal - to - noise ratios . the bright , unresolved seyfert nucleus of ngc 4151 creates a number of difficulties when trying to examine emission lines from the nlr close in . scattered light , largely from airy rings imaged on the slit , causes features of the nuclear point source spectrum to be superimposed on fainter nlr features . these follow linear tracks running nearly parallel to the dispersion , diverging slightly with wavelength and can be detected as much as 20@xmath1330 pixels from the nucleus ( bowers & baum , 1998 ) . this is a particularly difficult problem for measuring the balmer lines in the nlr since the blr lines are strong and often have peculiar shapes which can influence the continuum placement if not subtracted properly . in addition , the extended halo of the psf must be modeled and subtracted . furthermore , reflection of the bright nucleus in the ccd modes appears as a ghost spectrum , which is displaced from the nucleus in both the dispersion and spatial directions . several techniques were used to remove these effects . corrections for scattered light in the spectra were applied in the following order : 1 ) removal of the reflection spectrum ( in the ccd spectra ) , 2 ) correction for the halo , and 3 ) removal of the remaining psf , including the diffraction - ring tracks . the reflection spectrum is not only shifted in both directions , it is broadened in the spatial direction , compressed in the dispersion direction , and altered in intensity as a function of wavelength ( it tends to be redder than the nuclear spectrum ) . the reflection in each original spectral image was isolated by subtracting the scattered - light at the same spatial distances on the other side of the nuclear spectrum . then the nuclear spectrum was shifted along the slit and compressed in the dispersion direction until the strong emission features matched those in the reflection . it was then divided into the observed reflection to obtain the large - scale intensity variations in both dispersion and spatial directions . these variations were fitted in wavelength regions that do not contain extended emission , in both directions with low - order splines . the fits were then multiplied by the altered nuclear spectrum to produce a model of the reflection which was subtracted from the original spectral image . a circularly - symmetric halo was adopted from previous work on the stis detectors ( lindler 1999 ) , and collapsed to match the observed psf in the spatial direction ( obtained by adding regions along the dispersion direction that do not contain extended emission ) . the halo function was adjusted at various radial positions until a reasonable match was obtained with the broad - scale profile of the psf ( i.e. , ignoring diffraction tracks , etc . ) . the halo was then deconvolved from the original image using an iterative technique that removes flux from the halo and places it in the core . to remove the remaining scattered light , a scattering template was constructed using archival observations of stars observed with the same grating and slit width . first , the template spectrum was normalized in the dispersion direction by dividing through by the spectrum summed along the slit . next , the template was smoothed in the dispersion direction , using a median filter with a 50 pixel wide window . the nuclear spectrum of ngc 4151 was then multiplied into the template to simulate the scattered light spectrum . the scattering subtracted spectra are clean of broad line emission as close as 4 pixels from the nucleus . because the nuclear h@xmath14 line in the g750l spectrum at p.a . 221 is saturated , the true line profile is distorted and complicates construction of the scattering template . a substitute for the saturated profile was obtained from the g750 m short exposure in our slitless spectroscopy with good results . an alternative approach was also applied which used the structure along the slit in a continuum region of the ngc 4151 spectrum itself to form the model template . first , the entire image was normalized by dividing each row ( which lies along the dispersion direction ) by the summed nuclear spectrum from the central four rows . a spline ( typically of order 11 ) was then fitted along each row in regions that do not contain emission lines . thus the fit is a model of the scattering as a function of wavelength and position along the slit for a point source spectrum of constant flux per unit wavelength . this procedure was effective in modeling the diffraction tracks as well as the overall psf . the spline fits were then multipled by the nuclear spectrum at each spatial position , and subtracted from the reflection- and halo - corrected image to produce a final corrected image , which was used for subsequent analysis . the resulting spectra are shown in figure [ pa221 ] and figure [ pa70 ] . the corrections bring out the structure in the bright lines , and allow us to see fainter lines that are not evident in the original spectra . the correction process was not perfect , as evidenced by the faint structure seen in the regions above and below the strong nuclear lines , particularly h@xmath14 @xmath15 . however , these problems are minor , and the contaminating effects of nuclear absorption and emission were removed well enough for accurate measurement of the extended emission , even very close to the nucleus . although our primary interest is the nlr , the data set also contains high quality nuclear spectra of ngc 4151 at two epochs . monitoring campaigns have shown pronounced variability in both the nuclear continuum and blr emission ( robinson _ et al . _ , 1994 , et al . _ , 1996 , et al . _ , 1996 , kaspi _ et al . _ , 1996 , ulrich _ et al . _ , 1997 , weymann _ et al . _ , 1997 , peterson _ 1998 ) . over a time interval of 33 days ( jan . 8 , 1998 and feb . 10 , 1998 ; see table 1 ) the nuclear continuum dropped by 17% at 3050 and 10% at 6924 decreasing monotonically between these two wavelengths . this degree of variation is consistent with that reported in short timescale variability studies @xcite . the variation of the blr emission lines is less pronounced than that of the continuum , with h@xmath16 and h@xmath3 showing a decrease in flux , while the change in the h@xmath14 + n[ii ] line profile is more difficult to evaluate . the absorption lines in our far - uv spectrum are similar to those in the fos spectra published by weymann _ ( 1997 ) but is at too low a resolution for comparison to the high resolution ghrs spectrum . emission line fluxes and their errors were measured along the slit in each spectral range for a total of 45 emission lines . individual spectra were extracted from the longslit spectra by summing along the slit . the size of the extraction bins was dictated by the need for reasonably accurate fluxes for the he ii @xmath171640 and @xmath174686 lines , which were used for the reddening corrections in paper ii . experimentation revealed that bin lengths of [email protected] ( 4 ccd pixels , 8 mama pixels ) within the inner @xmath181@xmath19 and [email protected] outside this region would provide reasonable signal - to - noise ratios for these lines , and still isolate the emission - line clouds that we identified in our earlier papers . in some cases slightly different bin sizes were used to isolate individual clouds or to increase the signal - to - noise ratios . to measure the line fluxes , first a linear fit to the continuum adjacent to each line was subtracted . typically the continuum was very close to zero following removal of the scattered light , but continuum subtraction was helpful in regions of residual structure . next , the extreme ends of the red and blue wings of the line were marked and the total flux and centroid were computed between these two points . the uncertainties in the line fluxes were estimated using the error arrays for each spectrum produced by calstis and a propagation of errors analysis ( bevington , 1969 ) . for the blended lines of h@xmath14 and [ n@xmath20ii]@xmath216548 , 6584 , and [ s@xmath20ii ] @xmath22 6717 , 6731 , we used the [ o@xmath20iii]@xmath175007 line as a template to deblend the lines ( see crenshaw & peterson 1986 ) . this was superior to gaussian fitting since the emission line profiles are often complex . the results of the emission line flux measurements are presented in table 2 where the flux values are listed relative to h@xmath3 and the h@xmath3 flux is given at the bottom in units of @xmath23 ergs @xmath12 s@xmath0 @xmath0 . the errors obtained for each flux are given in parentheses . because of the failure of the far - uv spectrum at p.a . 70@xmath24 no g140l spectrum was obtained and so a reddening correction using the he ii lines was not possible . although dereddening using the balmer decrement is certainly valid , a reliable extrapolation from the red to the blue and near - uv lines is uncertain . we prefer , therefore , to continue the analysis without the corrected line ratios taking care that any possible effects of reddening are accounted for by other means . correction of the line ratios for reddening is an important step for a detailed photoionization model and is therefore presented in paper ii for the data at p.a . 221@xmath11 . to extract information on the multiple kinematic components the [ oiii ] @xmath25007 and [ oiii ] @xmath24959 lines were fitted independently with one to three gaussians . many slit extractions showed two components although in only a few cases was there compelling evidence for a third . only velocity components measured independently at both [ oiii ] @xmath24959 and [ oiii ] @xmath25007 are included . to test that each component represented the true kinematics of the gas , we compared the velocities obtained at both [ oiii ] @xmath25007 and [ oiii ] @xmath24959 . only those components with velocity difference in the two lines less than or equal to twice the mean difference for all points were retained . the procedure was then repeated . the first iteration removed velocity components that were wildly discordant , and therefore unlikely to be real , while the second gave us confidence that the remaining components are physically significant . from the difference in velocity between components extracted at [ oiii ] @xmath24959 and [ oiii ] @xmath25007 , we estimate the standard deviation of the velocities to be 30 km s@xmath0 . the results for each slit position are given in table 3a and 3b where the gaussian components are listed in order of increasing velocity . negative slit positions correspond to the sw region and positive slit positions correspond to the ne . figure [ oiiihb ] shows portions of the longslit spectra centered on the [ oiii ] @xmath25007 , [ oiii ] @xmath24959 and h@xmath3 emission lines for both slit positions , with the ne end of the slit at the top . the complex velocity structure that has been noted in both ground - based and hst studies ( _ e.g. _ schulz 1990 , kaiser _ et al . _ 1999 , hutchings _ et al . _ , 1999 ) is seen including line splitting at several positions along the slit . note that as a result of our scattering correction and psf subtraction we are able to probe the emission line kinematics to within [email protected] of the nucleus . included within our slits are four of the high velocity regions ( absolute value of projected velocity greater than 400 km s@xmath0 ) reported in hutchings _ et al . _ ( 1999 ) and 20 of the clouds identified in kaiser _ ( 1999 ) ( tables 4a , 4b ) . our agreement with hutchings velocities is reasonable , ranging from a difference of 6 km s@xmath0 for region n detected at slit p.a . 221@xmath24 , to 160 km s@xmath0 for region d. while some of the difference is undoubtedly due to measurement uncertainties , there may be real differences due to the portion of the high velocity regions which fall within our slit . there may also be some uncertainty due to confusion of spectral and spatial information in the slitless data . we see components of high velocity gas not specifically identified by hutchings _ ( 1999 ) on both sides of the bicone at both slit positions ( tables 3a , 3b ) . this gas corresponds to high velocity gas imaged by hutchings _ ( 1999 ) , but for which velocities were not previously measured . the high velocity components generally account for a small fraction of the total flux in the [ oiii ] emission lines , again in agreement with the findings of hutchings _ et al . _ we find more high velocity components in slit position p.a . 70@xmath11 , which is close to the radio ridge line , than we do in the p.a . 221@xmath24 slit . however , there is some high velocity gas not associated with the radio emission . comparison of our velocities with those reported by kaiser _ ( 1999 ) is more difficult , since in several cases they reported single velocities for clouds where we find multiple velocity components . furthermore , there are instances of extended clouds for which our slit does not sample the entire cloud . if we compare only velocities for clouds for which we find a single velocity component and average our velocities for clouds occurring in more than one extraction bin , we find the average difference in velocities is -18 km s@xmath0 + /- 94 km s@xmath0 , ( in the sense v(this paper ) - v(kaiser _ etal . . this difference and range is comparable to what was found for the high velocity clouds , and can be attributed to the same causes . figure [ oiiivels ] shows the velocities of the individual [ oiii ] components from the gaussian deblending . points along p.a . 221@xmath24 are marked as solid points and along p.a . 70@xmath24 as open symbols . the horizontal bars indicate the size of the extracted spectrum used for the measurement along the slit . vertical error bars are omitted since the velocity uncertainties are comparable to the size of the points on the diagram ( see section [ ana ] ) . a systemic velocity of 997 km s@xmath0 has been subtracted from the data . the solid and dashed lines show results expected for our simple models described below . the results follow the velocity distribution determined from the slitless spectroscopy of kaiser @xmath1 ( 1999 ) and the plot is similar to their figure 8 , though without the extreme high velocities . the velocities at large distances from the nucleus are consistent with the rotation of the galactic disk , while closer in the velocities are strongly blue shifted sw of the nucleus and strongly redshifted to the ne . to better understand the kinematics we consider two possibilities for the general form of the velocity field : radial outflow from the nucleus and expansion directed away from the radio axis . we adopt the basic conical geometry of the nlr of ngc 4151 as modeled by pedlar _ ( 1993 ) , with the radio jet pointing 40@xmath24 from the line of sight and projected onto the plane of the sky at a p.a . of 77@xmath11 . after consideration of the well - known geometry of the host galaxy ( simkin , 1975 , bosma , ekers , & lequeux , 1977 ) we require that the cone opening angle be wide enough to include our line of sight to the nucleus and also to intersect the disk of the host galaxy , since the extended narrow line region ( enlr ) kinematics follow the rotation of the disk . ( 1993 ) estimate the opening angle to be 130@xmath11 . however , boksenberg _ et al . _ ( 1995 ) argue that the nlr is density bounded and the ionized gas only partially fills the cone . therefore we choose a narrow vertex angle of 70@xmath24 which is a better match to the observed nlr structure . the models are drawn schematically in figure [ cones ] . these models are used to estimate the radial velocity as a function of projected distance from the nucleus for each slit position angle . our purpose is not to produce a detailed match to the observed velocities of each individual cloud , but to test two ideas about general form of the nlr kinematics . therefore , we assume that the interior of the cone is uniformly filled and note that the observed velocity distribution is not expected to be as smooth or complete as the model , reflecting the way in which the emission line clouds fill the cone . for both the radial outflow model and the jet expansion model we consider two cases , one in which the flow has a constant velocity and one in which the flow decelerates as it moves outward . we model this decelerating flow as a @xmath25 dependence where @xmath26 in the radial flow model is distance from the nucleus and @xmath26 in the jet expansion model is distance from the radio axis . this particular form of deceleration is chosen since it seems to represent the data best and is meant only to illustrate the effect . the results are plotted in figure [ rvmod ] for all four models in the form of a model longslit spectrum of a single emission line comparable to figure [ oiiihb ] . the slit was chosen to lie along p.a . 70@xmath24 and to have a slit width of [email protected] as in our stis observations . these simulated spectra were then deblended using two gaussians at each slit position in the same manner as the real data . the velocities for the decelerating models are shown in figure [ oiiivels ] as the dashed lines ( one for each gaussian component ) for the case of jet expansion and the solid lines for the case of radial flow . for the case of expansion away from the radio axis we expect both large positive and large negative velocities relative to the systemic velocity at any given position along the slit . in the case of radial outflow , however , large positive velocities and velocities much closer to the systemic velocity will be observed on one side of the slit while on the other side , large negative velocities and velocities near the systemic velocity are expected . since for ngc 4151 the far side of the sw cone lies close to the plane of the sky , the bulk of the flow is transverse to the line of sight , yielding radial velocities close to the systemic value , while the near side is much closer to the line of sight yielding large approaching radial velocities . similar considerations hold for the ne cone except that the near side of the cone lies in the plane of the sky and the far side yields the large receding velocities . we conclude that the radial outflow case gives a better match to the observed velocity distribution than the case of expansion away from the jet . in the case of a radially decelerating flow the overall envelope of the highest velocities decreases as one moves away from the nucleus much as seen in figure [ oiiivels ] . although the match is not perfect it seems to follow the trend of less extreme velocities as one moves along the slit . from these simple models we can not exclude the possibility of some motion perpendicular to the radio jet . however , it does seem likely that the flow is dominated by a radial outflow from the nucleus which slows with distance and that any contribution from expansion away from the jet is less significant . an understanding of the physical conditions in the nlr can be obtained by considering how various line strengths change as a function of distance from the nucleus and with respect to each other . in paper ii ( kraemer _ et al . _ 1999 ) a detailed photoionization model is developed using the emission line fluxes presented here . in the current paper we present a simpler analysis . the ratio of [ oiii ] @xmath25007 to h@xmath3 is well known to be sensitive to the ionization parameter @xmath27 , where @xmath28 is the rate at which ionizing photons are emitted , @xmath29 is the distance to the nucleus , @xmath30 is the electron density , and @xmath31 is the speed of light . figure [ o3hbrat ] shows the [ oiii ] @xmath25007 to h@xmath3 ratio as a function of distance along the slit for both position angles . we use ratios that have not been corrected for extinction since the lines are close in wavelength and are therefore rather insensitive to reddening . we see from the diagram that the line ratio decreases with distance in the inner 2@xmath4 and recovers somewhat at larger radii on the ne side ( positive x - axis ) . this trend was shown in kaiser _ ( 1999 ) from the slitless spectroscopy who suggest that this apparent change in the ionization parameter with distance reflects a decrease in density . apart from the increase in [ oiii ] @xmath25007 / h@xmath3 on the extreme ne side of p.a . 221@xmath24 , there is no significant difference in the ratio between the two position angles , suggesting that while the ionization state of the gas may change moving away from the nucleus , it generally does not change laterally , _ i.e. _ with distance from the radio axis . a similar trend is seen in the ratio of [ oiii ] @xmath25007 to [ oii ] @xmath23727 , which is also sensitive to the ionization parameter . figure [ o3o2rat ] plots the ratio versus distance for both slit positions . again , the line fluxes have not been corrected for extinction but the dust is most likely patchy ( see paper ii ) and so is not likely to influence the overall trend and merely adds scatter . support for this comes from the fact that the trend is largely symmetric about the nucleus indicating that no large scale dust lanes pass through our aperture . furthermore , for the slit extractions where both he ii lines used for the extinction corrections are present ( along p.a . 221@xmath24 ) , the largest change in the [ oiii]/[oii ] ratio from dereddening was a decrease of @xmath32% ( see paper ii ) . therefore , to the extent that the distribution of dust is comparable along each slit position , the conclusion of a decreasing [ oiii]/[oii ] ratio with distance is robust . the safest conclusion to draw from these diagrams is that the density falls off with distance as suggested by kaiser _ ( 1999 ) and confirmed in paper ii . in fact , in the inner clouds the high [ oiii]/[oii ] most likely results from collisional de - excitation of the o@xmath33 ions . although , these trends could naively be considered an indication of decreasing ionization parameter with distance from the nucleus , the more detailed investigation in paper ii suggests a more constant ionization parameter and a density which declines less rapidly than @xmath34 . in figure [ s2vdist ] the density sensitive ratio of [ sii ] @xmath26717 to [ sii ] @xmath26731 is plotted as a function of distance , again for both slit positions . judging by the size of the error bars , much of the scatter in the diagram is real suggesting that the gas is rather clumpy , with regions of higher and lower density at various points along the slit . there is also an interesting drop in the ratio very close to the nucleus particularly in the data from the p.a . 70@xmath24 slit position , suggesting an increase in density there . generally , the [ sii ] ratios appear to be larger farther out particularly along p.a . 221@xmath24 ( solid dots ) , indicating a decrease in density with radius , at least in the partially ionized zone . using the five level - atom program developed by shaw & dufour ( 1995 ) and assuming a temperature of 15000 @xmath24 k ( see below ) we find that the density of the inner nlr is roughly @xmath35 while in the outer nlr and enlr the density has dropped to @xmath36 . this agrees with the results of robinson _ et al . _ ( 1994 ) who found density decreasing with distance from the nucleus in ngc 4151 , with an overall nlr density of @xmath37 , and a density in the enlr of @xmath38 . this is also in agreement with the interpretation of the decline in [ oiii ] @xmath25007 /h@xmath3 as the result of a decrease in density . the [ oiii]@xmath39/[oiii]@xmath40 ratio is well - known to be sensitive to the temperature of the gas . figure [ o3o3vdist ] shows the [ oiii ] ratio as a function of distance along the slit . the use of this ratio to calculate the temperatures is only valid for densities up to @xmath41 @xmath42 at which point collisional de - excitation begins to have an influence on the line strengths ( osterbrock , 1974 ) . furthermore the [ sii ] densities can not be used since they reflect densities in the partially ionized zone . thus we use results from paper ii for the gas densities which indicate that in the o@xmath43 zone the densities are below @xmath44 . the results from the five - level atom program give temperatures in the range of 12000@xmath24 k 17000 @xmath24 k. based on figure [ o3o3vdist ] there appears to be a slight trend for a decreasing ratio ( increasing temperature ) with distance from the nucleus . this is difficult to confirm , however , since reddening may play a role , tending to increase the observed ratio . paper ii gives a more detailed analysis of the physical conditions along the slit . diagrams plotting one line ratio against another can be used to investigate the origin of the photoionizing continuum . by choosing line ratios which consist of lines which are close in wavelength we can significantly reduce the effects of reddening ( see _ e.g. _ veilleux and osterbrock , 1989 ) . in figure [ ratioplots ] a , b , and c , we present the optical emission line ratios [ s ii ] @xmath45 , [ n ii ] @xmath46 , [ o i ] @xmath47 , respectively , plotted against [ oiii ] @xmath25007 /h@xmath3 . in each diagram the solid line separates star - forming regions from agn and is taken from veilleux and osterbrock ( 1989 ) . the dashed line is the power - law photoionization model for solar abundance taken from ferland and netzer ( 1983 ) . the ionization parameter varies from @xmath48 to @xmath49 from lower right to upper left . we find that the ngc 4151 nlr clouds occupy compact regions on these diagrams indicating that the source of the ionizing continuum is the same for all of the points sampled along the slit . thus none of the clouds observed shows evidence for star - formation or liner - like excitation . while this result is not unexpected it is worth commenting that the nlr gas all seems to have the same source of excitation . other line ratio diagrams including uv lines are also interesting since they allow us to investigate the possibility of alternate ionization mechanisms for the nlr clouds ( allen _ et al . _ 1998 ) . in figure [ shockgrids]a , b and c we plot the ratios of civ @xmath50 to he ii @xmath51 , civ @xmath50 to ciii ] @xmath52 , [ ne v ] @xmath53 to [ ne iii ] @xmath54 , respectively against [ oiii ] @xmath25007 to h@xmath3 ( only the p.a . 221@xmath24 data is shown for figures [ shockgrids]a and b since the far - uv observation at p.a . 70 was unsuccessful ) . the lines show model grids calculated using the mappings ii code ( sutherland and dopita , 1993 ) by allen _ ( 1999 ) for shock ionization ( bottom ) , shock plus ionized precursor gas ( middle ) and for power - law photoionization ( top ) . for the shock plus precursor models , the shock velocity increases from 200 km s@xmath0 to 500 km s@xmath0 moving from low to high [ oiii ] @xmath25007 /h@xmath3 ratios . notice that for the highest velocity shocks the models coincide with power - law photoionization models . again the ngc 4151 nlr occupies very limited regions in these diagrams corresponding to photoionization by a power law at high ionization parameter or by shock plus precursor models with very high velocity ( @xmath55 km s@xmath0 ) . these results strongly suggest that low velocity shocks play an insignificant role in accounting for the ionization state of the nlr in ngc 4151 but we can not rule out the possibility of ionization by radiation from fast shocks . the results of the kinematic and emission line ratio analysis can be combined to create a coherent picture of the nlr in ngc 4151 . we have seen that the kinematics bear the signature of radial outflow from the nucleus and are distinctly different from an expansion away from the radio jet axis . this is an interesting result since many recent studies have reported kinematic evidence that the radio jet can have a significant influence on the motion of nlr gas ( _ e.g. _ bicknell _ et al . _ 1998 , ngc 4151 winge _ et al _ 1998 , mrk 3 capetti _ 1998 ) . in these studies the nlr gas is found immediately surrounding and expanding away from knots of radio emission as in mrk 3 or forms a bow shock structure around the working surface of the head of the jet as in mrk 573 ( capetti _ et al . _ 1996 , falcke , wilson , & simpson 1998 ) . this seems not to be the case for ngc 4151 . in conflict with this statement , the study of ngc 4151 by winge _ et al . _ ( 1998 ) reports that high velocity clouds are seen around the edges of the radio knots . this is not confirmed by kaiser _ ( 1999 ) who conclude that there is no direct association between non - virial gas kinematics , as determined by high velocity and high velocity dispersion , and proximity to the radio knots . our results concur with those of kaiser _ et al . _ ( 1999 ) . while we do find high velocity clouds in our aperture there is no distinct preference for them to found along p.a . 70@xmath11 , which is more closely aligned with the radio axis ( p.a . 77@xmath11 ) . further support for radial outflow comes from the emission line ratios as a function of position . for example there is no significant difference in the [ oiii]/[oii ] or [ oiii ] @xmath25007 /h@xmath3 ratios between the two slit positions even though the spectra at p.a . 70@xmath24 are much more closely aligned with the radio axis than the clouds at p.a . this is in contrast to the case of ngc 1068 where the [ oiii]/[oii ] ratio increases dramatically in regions that coincide with the radio jet ( axon _ et al . wfpc2 images of ngc 1068 presented by capetti , axon , & macchetto ( 1997 ) may also indicate higher density and ionization state along the radio jet in this object . furthermore , these authors suggest that an additional source of local ionizing continuum is required to explain the observations . while these results certainly raise an interesting possibility for ngc 1068 , our results for ngc 4151 show no such association between the radio morphology and the emission line ratios . thus the radio jet in ngc 4151 seems to have little influence on the ionization state of the gas . similar results are seen for the [ sii ] ratio and the [ oiii ] @xmath56 ratio suggesting no strong changes in the physical condition of the gas with proximity to the radio emission . because the line ratio diagrams show no evidence for shock or shock plus precursor ionization models at least for low velocity shocks , they support the arguments for radial outflow . if the gas were expanding away from the radio axis one would expect to see large amounts of shocked material particularly at the interface of the flow with the ambient interstellar medium of the host galaxy disk . in the case of radial outflow , we would expect to see little shocked gas since the motion is not directed into the disk and the relative velocities of gas within the flow should be small . perhaps an important consideration is that the radio morphology of ngc 4151 is rather different from that of mrk 3 for example . et al . _ have compared the radio structure of ngc 4151 to that of an fr i type radio galaxy , with much of the radio emission coming from a diffuse component , although on much smaller scales . the radio emission in mrk 3 , by contrast , is more jet - like being unresolved with merlin perpendicular to the radio source axis ( kukula _ et al . thus we might consider that the radio emission in ngc 4151 is not a well collimated jet , but rather a broad spray of plasma . gas clouds in the vicinity of the radio flow would thus be more naturally accelerated in directions roughly aligned with the radio axis than perpendicular to it . one possible scenario is that the core of the radio jet in ngc 4151 has cleared a channel in the line emitting gas and has blown out of the disk of the galaxy as suggested by schulz ( 1988 ) . thus there may have been a bow shock associated with the radio lobes in the past but the jet has passed on to lower density region in the outer bulge and galaxy halo . the line emitting gas is now free to flow out along the radio axis but only weakly interacts with the jet itself and the host galaxy ism . ngc 4151 is also known to have a system of nuclear absorption lines , particularly civ @xmath171549 , which are blueshifted with respect to the systemic velocity by values ranging from 0 to 1600 km s@xmath0 ( _ e.g. _ weymann _ et al . it is tempting to link the outflow seen in our study with that for the absorption line system . however , these flows are observed on vastly different scales and thus a true connection has not been established . models invoking winds from the nucleus to explain the nlr kinematics and other properties of seyfert galaxies have been proposed ( _ e.g. _ krolik & vrtilek , 1984 , schiano , 1986 , smith , 1993 ) . one suggestion is that x - ray heating of the molecular torus is the source of the wind ( krolik & begelman , 1986 ) . the base of the wind forms the electron scattering region which serves as the `` mirror '' allowing a view of the blr in polarized light in some seyfert 2 galaxies . at larger radii one might expect that the steep potential of the galaxy bulge tends to decelerate the wind . we conclude that the kinematics in ngc 4151 seem to be consistent with wind models for the nlr . the results presented in this paper provide an interesting contrast to the recent work on the nlr of seyfert galaxies . our analysis of the longslit spectra of ngc 4151 has revealed a rather different picture of the nlr in the sense that the prominent radio jet has very little influence on the kinematics and physical conditions . we find that the kinematics are best characterized by a decelerating radial outflow from the nucleus in the form of a wind . the lack of evidence for strong shocks near the radio axis and the uniformity of the line ratios across the nlr supports this picture . thus it appears that while interaction between the radio jet and the nlr gas may be a common occurrence it is by no means ubiquitous and does not apply in the case of ngc 4151 . we would like to thank diane eggers for her assistance in the data analysis . we would also like to thank mark allen for providing the model grids for the uv line ratio diagrams . this research has been supported in part by nasa under contract nas5 - 31231 . baum , s. , bohlin , r. , christensen , j. , debes . j. , downes , r. , ferguson , h. , gonnella , a. , goudfrooij , p. , hayes , j. , hodge , p. , hulbert , s. , katsanis , r. , keys , t. , lanning , h. , mcgrath , m. , sahu , k. , shaw , r. , smith , e. , walborn , n. , wilson , j. & bowers , c. 1998 stis instrument handbook , version 2.0 ( baltimore : stsci ) boksenberg , a. , catchpole , r. m. , macchetto , f. , albrecht , r. , barbieri , c. , blades , j. c. , crane , p. , deharveng , j. m. , disney , m. j. , jakobsen , p. , kampermann , t. m. , king , i. r. , mackay , c. d. , paresce , f. , weigelt , g. , baxter , d. , greenfield , p. , jedrzejewski , r. , nota , a. , & sparks , w. b. 1995 , apj , 440 , 151 hutchings , j.b . , crenshaw , d.m . , kaiser , m.e . , kraemer , s.b . , weistrop , d. , baum , s. , bowers , c.b . , feinberg , l. d. , green , r.f . , gull , t.r . , hartig , g.f . , hill , g. , lindler , d.j . , 1998 , apj , 492 , l115 llllllcrc + + + + & dataset & date & grating & dispersion & spectral & pa & ex . t. & + & & & & / pixel & range & ( @xmath11 ) & ( sec ) & + + & o42302070 & feb . 10 , 1998 & g430l & 2.73 & 2900@xmath135700 & 221 & 720 & + & o42302080 & feb . 10 , 1998 & g750l & 4.92 & 5240@xmath1310270 & 221 & 720 & + & o42302090 & feb . 10 , 1998 & g140l & 0.6 & 1150@xmath131730 & 221 & 2250 & + & o423020a0 & feb . 10 , 1998 & g140l & 0.6 & 1150@xmath131730 & 221 & 2250 & + & o42303050 & jan . 8 , 1998 & g430l & 2.73 & 2900@xmath135700 & 221 & 720 & + & o42303060 & jan . 8 , 1998 & g750l & 4.92 & 5240@xmath1310270 & 221 & 720 & + & o42303070 & jan . 8 , 1998 & g230 lb & 1.35 & 1680@xmath133060 & 221 & 2160 & + & o42304090 & june 1 , 1998 & g430l & 2.73 & 2900@xmath135700 & 70 & 720 & + & o423040a0 & june 1 , 1998 & g750l & 4.92 & 5240@xmath1310270 & 70 & 720 & + & o423040b0@xmath57 & jun . 1 , 1998 & g140l & 0.6 & 1150@xmath131730 & 70 & 2250 & + & o423040c0@xmath57 & jun . 1 , 1998 & g140l & 0.6 & 1150@xmath131730 & 70 & 2250 & + & o42305050 & may 7 , 1998 & g430l & 2.73 & 2900@xmath135700 & 70 & 720 & + & o42305060 & may 7 , 1998 & g750l & 4.92 & 5240@xmath1310270 & 70 & 720 & + & o42305070 & may 7 , 1998 & g230 lb & 1.35 & 1680@xmath133060 & 70 & 2160 & + & & & + + & + & & & + + + & & + & extraction & knot@xmath58 & region@xmath59 & extraction & knot@xmath58 & region@xmath59 + & @xmath60 & @xmath13 & @xmath13 & @xmath60 & @xmath13 & @xmath13 + & @xmath61 & 26 & n & @xmath61 & 1 & @xmath13 + & @xmath62 & 23 & @xmath13 & @xmath62 & @xmath13 & @xmath13 + & @xmath63 & @xmath13 & @xmath13 & @xmath63 & 10 & h + & @xmath64 & 22 & @xmath13 & @xmath64 & 10 & @xmath13 + & @xmath65 & @xmath13 & @xmath13 & @xmath66 & 4 & @xmath13 + & @xmath67 & @xmath13 & @xmath13 & @xmath68 & 4 & @xmath13 + & @xmath69 & 25 & @xmath13 & @xmath70 & 4 & @xmath13 + & @xmath66 & 25 & @xmath13 & @xmath71 & 5 & @xmath13 + & @xmath68 & 25 & @xmath13 & @xmath72 & 5 & @xmath13 + & @xmath70 & 25 & @xmath13 & @xmath73 & @xmath13 & @xmath13 + & @xmath71 & @xmath13 & @xmath13 & @xmath74 & @xmath13 & @xmath13 + & @xmath72 & @xmath13 & @xmath13 + & @xmath75 & @xmath13 & @xmath13 + & @xmath76 & @xmath13 & @xmath13 + & @xmath77 & @xmath13 & @xmath13 + & @xmath78 & @xmath13 & @xmath13 + & @xmath79 & 33 & @xmath13 + & @xmath80 & 37 & @xmath13 + llcclcc + + & & + & extraction & knot@xmath58 & region@xmath59 & extraction & knot@xmath58 & region@xmath59 + & @xmath60 & @xmath13 & @xmath13 & @xmath60 & @xmath13 & @xmath13 + & @xmath61 & 26 & n & @xmath61 & 11,6 & @xmath13 + & @xmath62 & 23 & @xmath13 & @xmath81 & 6 & @xmath13 + & @xmath63 & 19 & @xmath13 & @xmath82 & 9,12 & @xmath13 + & @xmath64 & @xmath13 & d , d & @xmath83 & 12 & @xmath13 + & @xmath84 & 16 & @xmath13 & @xmath85 & @xmath13 & @xmath13 + & @xmath86 & 16 & @xmath13 & @xmath87 & @xmath13 & @xmath13 + & @xmath88 & @xmath13 & @xmath13 & @xmath89 & 3 & @xmath13 + & @xmath90 & @xmath13 & @xmath13 & @xmath91 & 3 & @xmath13 + & @xmath92 & @xmath13 & @xmath13 & @xmath72 & 14 & @xmath13 + & @xmath75 & 18 & @xmath13 & @xmath75 & @xmath13 & @xmath13 + & @xmath76 & 18 & @xmath13 & @xmath76 & @xmath13 & @xmath13 + & & & & @xmath74 & @xmath13 & @xmath13 + & & & & @xmath93 & @xmath13 & @xmath13 + & & & & @xmath94 & @xmath13 & @xmath13 + & & & & @xmath95 & 29 & @xmath13 +	longslit spectra of the seyfert galaxy ngc 4151 from the uv to near infrared have been obtained with stis to study the kinematics and physical conditions in the nlr . the kinematics show evidence for three components , a low velocity system in normal disk rotation , a high velocity system in radial outflow at a few hundred km s@xmath0 relative to the systemic velocity and an additional high velocity system also in outflow with velocities up to 1400 km s@xmath0 , in agreement with results from stis slitless spectroscopy ( hutchings _ et al . _ , 1998 , kaiser @xmath1 , 1999 , hutchings _ et al . _ , 1999 ) we have explored two simple kinematic models and suggest that radial outflow in the form of a wind is the most likely explanation . we also present evidence indicating that the wind may be decelerating with distance from the nucleus . we find that the emission line ratios along our slits are all entirely consistent with photoionization from the nuclear continuum source . a decrease in the [ oiii ] @xmath25007 / h@xmath3 and [ oiii ] @xmath25007 / [ oii ] @xmath23727 ratios suggests that the density decreases with distance from the nucleus . this trend is borne out by the [ sii ] ratios as well . we find no strong evidence for interaction between the radio jet and the nlr gas in either the kinematics or the emission line ratios in agreement with the results of kaiser _ et al . _ ( 1999 ) who find no spatial coincidence of nlr clouds and knots in the radio jet . these results are in contrast to other recent studies of nearby agn which find evidence for significant interaction between the radio source and the nlr gas .
the outstanding properties of whispering - gallery mode resonators ( wgmrs ) @xcite have enabled breakthroughs in various fields such as single - particle sensing @xcite , coupling to single atoms @xcite , narrow - band optical filtering @xcite and lasing @xcite , optomechanical effects @xcite , and wgmr - enhanced nonlinear optics @xcite being the basis for frequency comb generation @xcite and quantum optics @xcite . wgmrs support a discrete set of eigenmodes , the so called whispering - gallery modes ( wgms ) . most wgmr experiments could benefit from the knowledge of mode numbers for an exact quantification of their parameters . in sensing experiments with wgmrs , for instance , exact information on the spatial overlap of the wgm and the probe particle is advantageous @xcite . in nonlinear optics , information on spectral properties , in addition to the spatial overlaps is needed to address and quantify conversion channels @xcite . there are straight forward analytical @xcite and numerical methods to calculate the mode structure in wgmrs that are only slightly larger than the wavelength . numerical studies in large wgmrs , however , pose a significant problem as even with today s computational possibilities full three - dimensional modeling is extremely difficult . a complete description for the far - field emission patterns from spheroidal wgmrs , as being experimentally investigated in this work , is still an open question in the field . in addition , the high number of experimentally accessible modes in large wgmrs makes mode analysis a very challenging task . several experimental methods for mode analysis have been studied in the past , such as near - field probing @xcite , far field imaging @xcite and investigating the spectral response of these resonators @xcite . regarding the breakthrough experiments mentioned earlier in the text , an experimental characterization of a macroscopic - size wgmr mode structure has only been achieved in the context of optical sum - frequency generation @xcite on the basis of sideband spectroscopy @xcite . this technique requires an optical probe coupled to the mode of interest at a wavelength that may not be available . furthermore , finding the mode of interest , e.g. the signal or idler down - converted mode , with the probe laser is experimentally challenging . the spatial structure and the resonance frequency of each wgm is characterized by a unique set of numbers , i.e. the polar , the azimuthal , and the radial mode number @xmath0 , m , and q , respectively . the full information on these mode numbers is in principle contained in the far - field images of the outcoupled modes @xcite , and in the frequency spectrum @xcite . the exact identification of wgms can be extremely difficult in practice using only one of these approaches . here we present a combination of these techniques that allows for a complete identification of all significantly coupled wgms of a large scale ( millimeter - sized ) wgmr . first , we determine the radial mode numbers q and the angular mode numbers p=@xmath0-m via an analysis of the observed far - field emission patterns . then knowing these two numbers we unambiguously find the polar mode numbers @xmath0 by fitting the frequency spectrum . this fitting results in a very accurate determination of the product of the refractive index @xmath3 times the major radius @xmath4 of the resonator . prior knowledge of either @xmath3 or @xmath4 allows the determination of the other parameter . this detailed characterization of the properties of the wgmr and the mode structure is important for wgmrs to become a versatile standard for classical and quantum optics . the manuscript is structured as follows : section one theoretically describes far - field emission patterns based on the electric field distributions of wgms . section two explains the mode spectrum of wgmrs via the dispersion relation . after a description of the experimental setup in section three , experimental results on the characterization of the wgmr , in particular mode analysis , are presented on the basis of measured emission patterns and frequency spectra in section four . far - field emission patterns from wgmrs carry information on the internal mode structure . according to maxwell s equations , wgms in spherically - symmetric resonators are described by spherical harmonic functions and spherical bessel functions @xcite . for wgms close to the equatorial plane and the surface of a spheroidal wgmr , the electric field amplitudes are approximated by @xcite ( see fig . [ fig : spatialwgms ] ) : @xmath5 [ eq : spatwgmmain ] where @xmath4 is the major radius of the wgmr and @xmath6 the curvature . analogous to the solution of maxwell s equations in a box with dirichlet boundary conditions , the wgms in this spheroidal geometry are characterized by three integer mode numbers : the polar , the azimuthal , and the radial mode numbers @xmath7 and @xmath8 , respectively ( see fig . [ fig : spatialwgms ] ) . the angular mode number @xmath9 of the wgms gives the degree of the hermite polynomials and therefore the number of field oscillations in @xmath10-direction . @xmath11 is the angular size parameter . the radial part of the electric field in eq . is described by the radial size parameter @xmath12 and the q - th root @xmath13 of the airy - function @xmath14 ) . of the observation point to the surface of the wgmr , the polar angle @xmath10 , and the azimuthal angle @xmath15 . the intensity distributions on the right - hand side are shown for various radial mode numbers q and angular mode numbers p=@xmath0-m . ] the spatial structure of a wgm with open boundaries , where evanescent fields are present in the close vicinity of the wgmr , can be probed by placing a prism with refractive index @xmath16 next to the wgmr . the far - field emission pattern from such a prism is described as a fourier transform of the evanescent wgm field at the prism surface . this near - field at the prism surface can be derived by applying a coupling window @xcite , i.e. a finite aperture , to the angular part of eq . . this operation is equivalent to a low - pass filtering for the far - field . according to eq . , the mode profile in @xmath10-direction contains information on the angular mode number p. for equatorial wgms ( p=0 ) ( see fig . [ fig : spatialwgms ] ) , the near - fields and therefore the far - fields show exactly one maximum . for @xmath17 , the far - fields in @xmath10-direction show two distinct lobes symmetric around the equatorial plane for large spheres @xcite . this is not generally the case for aspherical resonators . for oblate resonators ( @xmath18 ) , these far - field distributions can also be modeled with two main maxima that are symmetric about this plane . according to the experimental results for our resonator ( @xmath19 ) and p=2 , these maxima overlap strongly ( see section three for experimental results ) . for p=4 , they can be resolved . for prolate resonators ( @xmath20 ) , the effect of the coupling window is reduced . in the extreme case of @xmath21 , the width of the wgm in the near - field is much smaller than the width of the coupling window . hence , the coupling window does not affect the out - coupling and more detailed features of the respective wgm mode structure are visible in the far - field . for equatorial wgms , the equatorial coupling angle @xmath22 and the divergence @xmath23 are given by @xcite : @xmath24 the equatorial angle @xmath22 is in good approximation equivalent to the critical angle of total internal reflection of the resonator and the prism material and is independent of the minor radius @xmath6 . a measurement of the equatorial angle @xmath22 reveals the polar mode number @xmath0 , which is equal to the azimuthal mode number m for equatorial modes . an understanding of the mode spectrum of wgmrs requires a discussion of the dispersion relation @xcite on the basis of frequency differences between individual wgms . using the dispersion relation @xcite , we connect the mode numbers @xmath0 , q , and @xmath25 to the resonant optical frequency @xmath26 : @xmath27 , \label{eq : disprel}\end{aligned}\ ] ] where p and @xmath28 are of order one . depending on the polarization , the parameter @xmath29 is 1 for te modes and @xmath30 for tm modes . the wavelength - dependent refractive index of the wgmr is @xmath3 . this renders the right - hand side of eq . frequency dependent.the q - th root of the airy function @xmath31 can be approximated as @xmath32^{2/3}$].the scaling factor @xmath33 appears as a refractive index dependent and thus frequency - dependent normalization factor . the spectrum around a certain wgm at frequency @xmath34 is found by evaluating frequency differences : @xmath35 to other wgms at frequencies @xmath36 . due to the material dispersion , eq . is an implicit function of the frequency . hence , the correct description of the frequency differences requires a full numerical approach . for a better qualitative understanding of the relevant quantities for the formation of the frequency spectrum , we can give analytic expressions using first order approximations . for this , we define a dispersive scaling factor @xmath37 by including the slope of the refractive index @xmath38 at the frequency @xmath34 as : @xmath39 small contribution of the polarization dependent term @xmath40 in eq . are omitted . in the following , we model the frequency spectrum on the basis of changes in the respective mode numbers @xmath41 , and @xmath42 . this discussion is illustrated in fig . [ fig : fsrmodespacing ] . the free spectral ranges @xmath43 originate from steps in the @xmath0 number and depend only on @xmath0 and q : @xmath44 . \label{eq : fsrdip}\end{aligned}\ ] ] for increasing polar mode numbers @xmath0 , found for example at higher optical frequencies , wgms are located closer to the wgmr surface . this increases the effective radius for these wgms and decreases the @xmath45 . in contrast , an increase in the radial mode number q leads to the field shifting away from the surface , which decreases the effective radius and increases the @xmath43 . , see eq . ) are determined by geometric dispersion . in addition to equatorial wgms @xmath46 , each mode family contains non - equatorial wgms @xmath47 at an offset frequency given by eq . . within a given spectral observation window , the frequency differences of the wgms belonging to different mode families allow to unambiguously identify the polar mode number @xmath0 . ] the spacings between different radial wgms : @xmath48 can exceed the @xmath43 by orders of magnitude and depend only on @xmath0 and q. within the spectral observation window , modes with a different radial number q will also have a different polar number @xmath49 ( see fig . [ fig : fsrmodespacing ] ) . the spacings between wgms with different angular mode numbers p : @xmath50 are determined solely by the ratio of the radii @xmath51 and independent on q and @xmath0 . eq . has been used extensively in the context of frequency comb generation @xcite . for @xmath52 where @xmath53 , the spacings can match multiples of the @xmath43 given by eq . , which can lead to degenerate frequencies @xmath54 . the case of a spherical resonator ( n=1 ) , which exhibit degenerate mode families , is well known . according to eq . , spheroidal resonators may have sufficient symmetry to support degenerate mode families . the frequency spectrum depends on a variety of parameters of the wgmr and the environment , such as temperature , material , pressure , and geometry . each of these parameters can be used independently to tailor the frequency spectrum . in principle , accurate knowledge about the scaling factor @xmath55 , i.e. the fundamental value @xmath56 , together with an absolute frequency measurement of one wgm can provide a way to determine mode numbers @xmath0 , q , and p of this particular wgm . this can be experimentally challenging due to the limited knowledge of the geometry and the refractive index of the wgmr . in contrast , the experimental study presented in the following two sections is based on relative frequency measurements . first , we obtain the mode numbers q and p from a spatial mode analysis . using this knowledge , we show a complementary analysis of the frequency differences @xmath57 from the fundamental wgm to higher - order wgms to obtain the wgmr properties , such as the scaling factor @xmath55 , the major radius @xmath4 , and minor radius @xmath6 . in this analysis , @xmath0 is exactly determined by a fitting procedure . the experimental setup shown in fig . [ fig : setup ] allows spatial and spectral characterization of wgms . the macroscopic wgmr is manufactured from a congruent 5.3% mgo - doped lithium niobate wafer , such that the optic axis is aligned with the symmetry axis of the resonator ( z - cut @xcite ) . the major and the minor radius of the disk are measured to be @xmath58 mm and @xmath59 mm with a microscope leading to a scaling factor of @xmath60 . the wgmr is mounted in an oven whose temperature is stabilized to a millikelvin level . the light source for probing the whispering - gallery resonator is a continuous wave laser at @xmath61 nm ( nd : yag prometheus , innolight ) with a hermite - gauss tem@xmath62 mode . its polarization is parallel to the optic axis of the wgmr ( te mode ) . evanescent coupling to the disk is achieved by focusing the beam onto the inner surface of a diamond prism under an angle of total internal reflection ( see fig . [ fig : setup ] ) . the transverse coupling angle was adjusted by maximizing coupling to equatorial modes . all light coming from the coupling point is collimated and sent to a photo detector . at the resonance frequencies of individual wgms , light can be coupled to the wgmr . tuning the laser frequency over more than an fsr , this configuration can be used for the spectral analysis . at the in - coupling prism , the emission pattern from the wgmr destructively interferes with the directly reflected spatial mode of the laser source . we use a second diamond prism for out - coupling to investigate the modes independently from this interference . the light emerging from the second prism is sent to a ccd camera placed at a distance of 1.4 cm from the coupling point , which gives us in a good approximation the far - field emission pattern . in our mode analysis technique , we combine the information from far - field emission patterns and from the frequency spectrum . in the first subsection mode numbers q and p are obtained from an analysis of the far - field emission patterns . in the second subsection a complementary analysis of the frequency differences @xmath57 from the fundamental wgm to higher - order wgms is performed to obtain the wgmr properties and the polar mode numbers @xmath0 . as a first step , we tune the pump laser frequency over 14 ghz and record the wgmr spectrum with a photodetector . it is known from sensing with wgmrs that objects within the evanescent fields , in this case the coupling prism , can induce a shift of the resonance frequencies @xcite . to avoid this effect , the experiment is carried out in the under - coupled regime of the fundamental wgm . the measured frequency spectrum is shown in fig . [ fig : atlas](a ) . the numbers @xmath63 of the modes shown in this figure are not yet determined . to find these numbers , we take far - field images ( see fig . [ fig : camerammfit](a ) ) of all wgms with a reasonably strong coupling within one fsr . with our coupling optimized for equatorial modes using a tem@xmath62 laser beam , the excitation of modes with odd p is strongly suppressed ( see also @xcite ) . consequently , we expect to see only even numbers of p , whose coupling decreases rapidly for higher p. note that tilting the pump beam from the equatorial plane excites an increasing number of non - equatorial modes . the spectrum becomes more dense and less pronounced features of well - coupled modes are visible . wgms with different p numbers are distinguished based on the cross section of the far - field emission patterns , which is depicted in fig . [ fig : camerammfit](b ) as an average for wgms with the same p but different q. equatorial wgms with p=0 show a single - lobe nearly gaussian emission pattern . wgms with p=2 exhibit a flat - top angular distribution . this distinction is clear enough to identify the p=0 and p=2 modes in most relevant cases . for wgms with p=4 , a distinct two - lobe structure appears . a limitation arises for strongly oblate ( @xmath64 ) resonators when emission patterns coming from p=0 and p=2 wgms strongly overlap . the measured emission patterns of the wgms were fitted with gaussian functions in horizontal direction . this reveals the information on the central positions ( see in fig . [ fig : camerammfit](c ) for the associated coupling angles @xmath22 ) and @xmath65e - values of the gaussian functions , and hence the divergences of the emission patterns . the central position give the information on the q number . since the resonance frequencies @xmath34 of all wgms within the spectral observation window ( see fig . [ fig : fsrmodespacing ] ) are approximately equal to the pump laser frequency , eq . implies that larger radial mode numbers q correspond to smaller numbers @xmath0 . hence , an increase in the radial mode number q results in a decrease in the out - coupling angle @xmath22 for equatorial wgms according to eq . . this corresponds to the horizontal shift of the center of the spatial profiles depicted in fig . [ fig : pics](a ) , which provides the basis for identification of q. this also means that coupling of modes with different q can be optimized via the coupling angle @xmath22 , which is observed in the experiment . nm of the macroscopic wgmr ( major radius @xmath66 mm ) measured at @xmath67c ( a ) and @xmath68c ( c ) . the radial mode numbers q and the angular mode numbers @xmath25 are assigned according to the far - field emission patterns ( see fig . [ fig : pics ] ) . frequencies are specified relative to the frequency @xmath69 of the fundamental wgm @xmath70 , which is highlighted in grey . non - equatorial wgms @xmath47 exhibit a fixed frequency difference to the respective equatorial wgm @xmath71 ( see fig . ( a ) and eq . ) . ( b ) shows measured ( square markers ) and calculated ( vertical lines ) frequency differences @xmath72 according to eq . for different equatorial wgms @xmath71 at different temperatures . temperature tuning of these wgms exhibits different slopes due to their different @xmath73 given by eq . . the frequency spectrum at a given temperature is then directly related to the polar mode number @xmath0 ( @xmath74 for equatorial wgms ) of the fundamental wgm shown on the right axis . ] therefore the analysis of the far - field emission patterns has allowed an assignment of radial mode number q and angular mode number p in the mode spectrum shown in fig . [ fig : atlas ] . the polar mode numbers @xmath0 are obtained from an analysis of frequency differences in the mode spectrum , which is discussed below . with the information on @xmath0 , q , and p , one can calculate the coupling angles @xmath22 using eq . . the calculated angles can be compared with the coupling angles inferred from geometry of the setup and the measured central positions of the far - field images . a constant offset angle is taken as a fitting parameter . the experimental results shown in fig . [ fig : camerammfit ] are in good agreement with the model . multiple measurements of one mode were taken to estimate the standard deviations for the measured coupling angles , which are presented as error bars in fig . [ fig : pics](c ) . for high radial mode numbers q , the error bars of adjacent wgms will start to overlap . adjacent wgms can then no longer be distinguished unambiguously due to statistical variation in the measured central positions . our assignment of radial mode numbers for equatorial wgms up to q = 12 is also confirmed by the frequency analysis described below . ( a ) for the corresponding frequency spectrum ) . the wgms \{q=2,p=0 } and \{3,2 } exhibit distortions in the far - field images , which probably originate from their frequency degeneracy to other wgms . the horizontal center positions of the intensity distributions shift to the left ( towards smaller @xmath22 ) for higher radial mode numbers q. b ) averaged vertical cross section for wgms with the same p but different q. this gives information on the distribution of the wgms in @xmath10-direction , which allows a distinction between different angular mode numbers p. c ) distinction of different radial mode numbers q of equatorial wgms ( p=0 ) according to the coupling angle @xmath22 . within our spectral observation window , the higher the radial mode number q , the smaller is @xmath0 and @xmath22 . the exact polar mode numbers @xmath0 used for the calculated out - coupling angles @xmath22 ( see eq . ) are derived from an analysis of the spectrum . ] [ fig : camerammfit ] the experimental @xmath65e - values of the mode profiles in @xmath75direction can now be translated into angular divergences . the mean value of the measured divergences @xmath76 matches well ( within @xmath77 ) with the theoretical value @xmath78 ( see eq . ) . the divergences are approximately equal for all measured radial modes . we use our knowledge of the mode numbers q and p for the spectral analysis . the frequency of the fundamental wgm \{q=1 , p=0 } is taken as a reference point . in fig . [ fig : atlas ] , all other modes were plotted relative to this fundamental wgm frequency . the measured frequency differences of the equatorial modes are fitted with calculated frequency differences according to the dispersion relation given by eq . . for p=0 , the dispersion relation depends on the radius @xmath4 and polar mode number @xmath0 as free parameters . a@xmath79 the mismatch between measured and computed frequency differences ( see eq . ) over a broad range of @xmath0 numbers of the fundamental wgm \{q=1,p=0}. the frequency mismatches of a selection of equatorial wgms are shown . the individual mismatches reach zero at multiple values of @xmath0 , however , all together they reach zero at the unique value of @xmath0 = 41903 in b ) . ] this fit ( see fig . [ fig : accuracy ] ) unambiguously yields the polar mode number @xmath0=41903 for the fundamental wgm , and confirms the assignment of q and p for the other wgms gained from the spatial analysis . for p=0 , the radius only appears in the scaling factor @xmath55 in the dispersion relation . therefore the fitted radius @xmath4 can only be trusted within the accuracy for the absolute value for measured frequency differences . a much better accuracy is achieved by evaluating the fitted polar mode number together with absolute frequency of the pump laser . taking the measured frequency @xmath80thz of the fundamental wgm determined with a wavemeter , the fit results in an optical circumference of @xmath81 . using the refractive index found from the sellmeier equation for lithium niobate @xcite , @xmath82 , we estimate the large radius to @xmath83 mm at 70@xmath84c . this is consistent with the value measured with the microscope but has a precision improved by an order of magnitude . additionally , knowledge of the mode numbers and the frequency spacing between non - equatorial modes described by eq . can be used to fit the ratio @xmath85 . together with the knowledge of @xmath4 , the fit yields @xmath86 mm . the measurements of the radii also agree well with the microscopically measured value . thermal expansion does not affect the wgmr radii within the accuracy of the microscope measurement . in a final step , the spectral characterization is carried out again at a different temperature of the wgm resonator within the same spectral observation window ( see fig . [ fig : atlas](c ) ) . this results in a shift of the relative frequencies for all the modes within the spectrum ( see fig . [ fig : atlas](b ) ) . a change in temperature effectively changes the resonator optical length , both through thermal expansion and the thermal change in the refractive index of lithium niobate . this changes the size parameter @xmath55 of the resonator . the change in the measured mode spectrum in the fixed spectral observation window can thus be directly related to a change in the @xmath0 number for each mode in the observation window by @xmath87 . in principle , temperature tuning can also be used to obtain accurate information on the thermo - refractive and the thermal expansion coefficient . due to experimental uncertainties in the temperature determination , this is not possible here with a reasonable accuracy . in contrast , we use temperature tuning to demonstrate an efficient method for tailoring the frequency spectrum by choosing the proper @xmath0 number at a certain temperature . in summary , we have demonstrated a practical experimental technique to identify mode numbers of all significantly coupled wgms within the spectral observation window by evaluating far - field emission patterns and frequency differences . this technique takes advantage of geometrical dispersion arising in wgms due to their curvature . therefore it is more efficient in application to small resonators . however , we have shown that the exact determination of the mode numbers is possible even in relatively large resonators . apart from analyzing an unknown mode spectrum , the understanding of the frequency spectra can also be used to specifically tailor wgm spectra . in addition , this measurement technique allows for accurate measurements of the resonator radii @xmath4 and @xmath6 times refractive index @xmath3 . the determination of the optical circumference @xmath88 is mainly limited by the uncertainty of the laser frequency . this can be increased experimentally by orders of magnitude to allow very accurate measurements of either resonator geometry or refractive index . the large set of accurately characterized wgms can be employed to select wgms with specific spatial or spectral properties on demand for a huge variety of applications in classical and quantum optics . the authors are grateful for the financial support of the european research council under the advanced grant pacart . finally , the authours would like to thank peter banzer , thomas bauer and martin neugebauer for fruitful discussions .	identifying the mode numbers in whispering - gallery mode resonators ( wgmrs ) is important for tailoring them to experimental needs . here we report on a novel experimental mode analysis technique based on the combination of frequency analysis and far - field imaging for high mode numbers of large wgmrs . the radial mode numbers q and the angular mode numbers p=@xmath0-m are identified and labeled via far - field imaging . the polar mode numbers @xmath0 are determined unambiguously by fitting the frequency differences between individual whispering gallery modes ( wgms ) . this allows for the accurate determination of the geometry and the refractive index at different temperatures of the wgmr . for future applications in classical and quantum optics , this mode analysis enables one to control the narrow - band phase - matching conditions in nonlinear processes such as second - harmonic generation or parametric down - conversion . 10 a. matsko and v. ilchenko , `` optical resonators with whispering - gallery modes - part i : basics , '' ieee j. sel . top . quantum . electron . * 12 * , 314 ( 2006 ) . v. braginsky , m. gorodetsky , and v. ilchenko , `` quality - factor and nonlinear properties of optical whispering - gallery modes , '' phys . lett . a * 137 * , 393397 ( 1989 ) . f. vollmer and s. arnold , `` whispering - gallery - mode biosensing : label - free detection down to single molecules , '' nat . methods * 5 * , 591596 ( 2008 ) . f. sedlmeir , r. zeltner , g. leuchs , and h. g. l. schwefel , `` high - q mgf@xmath1 whispering gallery mode resonators for refractometric sensing in aqueous environment , '' arxiv:1409.2539 [ physics ] ( 2014 ) . d. vernooy , a. furusawa , n. georgiades , v. ilchenko , and h. kimble , `` cavity qed with high - q whispering gallery modes , '' phys . rev . a * 57 * , 22932296 ( 1998 ) . t. aoki , b. dayan , e. wilcut , w. p. bowen , a. s. parkins , t. j. kippenberg , k. j. vahala , and h. j. kimble , `` observation of strong coupling between one atom and a monolithic microresonator , '' nature * 443 * , 671674 ( 2006 ) . s. p. vyatchanin , m. l. gorodetsky , and v. s. ilchenko , `` tunable narrow - band optical filters with modes of the whispering gallery type , '' j. appl . spectrosc . * 56 * , 182187 ( 1992 ) . m. c. collodo , f. sedlmeir , b. sprenger , s. svitlov , l. j. wang , and h. g. l. schwefel , `` sub - khz lasing of a caf2 whispering gallery mode resonator stabilized fiber ring laser , '' opt . express * 22 * , 19277-19283 ( 2014 ) . j. hofer , a. schliesser , and t. j. kippenberg , `` cavity optomechanics with whispering - gallery mode optical micro - resonators , '' adv . atom . mol . . phy . * 58 * , 207323 ( 2010 ) . t. herr , v. brasch , j. d. jost , c. y. wang , n. m. kondratiev , m. l. gorodetsky , and t. j. kippenberg , `` temporal solitons in optical microresonators , '' nat . photonics * 8 * , 145152 ( 2013 ) . j. yu , k. giulietti , f. sourgen , a. ross , j. p. wolf , m. ferriol , g. foulon , c. goutaudier , m. t. cohen - adad , and g. boulon , `` second - harmonic generation in a microradius linbo@xmath2 cylinder with a quasi - elliptical cross section , '' opt . lett . * 24 * , 394396 ( 1999 ) . a. savchenkov , a. matsko , d. strekalov , m. mohageg , v. ilchenko , and l. maleki , `` low threshold optical oscillations in a whispering gallery mode caf@xmath1 resonator , '' phys . rev . lett . * 93 * , 243905 ( 2004 ) . j. frst , d. v. strekalov , d. elser , m. lassen , u. l. andersen , c. marquardt , and g. leuchs , `` naturally phase - matched second - harmonic generation in a whispering - gallery - mode resonator , '' phys . rev . lett . * 104 * , 153901 ( 2010 ) . j. frst , d. strekalov , d. elser , a. aiello , u. andersen , c. marquardt , and g. leuchs , `` low - threshold optical parametric oscillations in a whispering gallery mode resonator , '' phys . rev . lett . * 105 * , 263904 ( 2010 ) . c. marquardt , d. strekalov , j. frst , m. frtsch , and g. leuchs , `` nonlinear optics in crystalline whispering gallery resonators , '' opt . photon . news * 24 * , 38 - 45 ( 2013 ) . d. v. strekalov , a. s. kowligy , y .- p . huang , and p. kumar , `` optical sum - frequency generation in a whispering - gallery - mode resonator , '' new j. phys . * 16 * , 053025 ( 2014 ) . p. delhaye , a. schliesser , o. arcizet , t. wilken , r. holzwarth , and t. j. kippenberg , `` optical frequency comb generation from a monolithic microresonator , '' nature * 450 * , 12141217 ( 2007 ) . w. liang , a. a. savchenkov , a. b. matsko , v. s. ilchenko , d. seidel , and l. maleki , `` generation of near - infrared frequency combs from a mgf@xmath1 whispering gallery mode resonator , '' opt . lett . * 36 * , 22902292 ( 2011 ) . j. frst , d. strekalov , d. elser , a. aiello , u. andersen , c. marquardt , and g. leuchs , `` quantum light from a whispering - gallery - mode disk resonator , '' phys . rev . lett . * 106 * , 113901 ( 2011 ) . m. frtsch , j. u. frst , c. wittmann , d. strekalov , a. aiello , m. v. chekhova , c. silberhorn , g. leuchs , and c. marquardt , `` a versatile source of single photons for quantum information processing , '' nat . commun . * 4 * , 1818 ( 2013 ) . c. s. werner , t. beckmann , k. buse , and i. breunig , `` blue - pumped whispering gallery optical parametric oscillator . '' . lett . * 37 * , 42244226 ( 2012 ) . c. c. lam , p. t. leung , and k. young , `` explicit asymptotic formulas for the positions , widths , and strengths of resonances in mie scattering , '' j. opt . soc . am . b * 9 * , 15851592 ( 1992 ) . j. c. knight , n. dubreuil , v. sandoghdar , j. hare , v. lefvre - seguin , j. m. raimond , and s. haroche , `` mapping whispering - gallery modes in microspheres with a near - field probe , '' opt . lett . * 20 * , 15151517 ( 1995 ) . a. mazzei , s. gtzinger , l. s. menezes , v. sandoghdar , and o. benson , `` optimization of prism coupling to high - q modes in a microsphere resonator using a near - field probe , '' opt . commun . * 250 * , 428433 ( 2005 ) . a. savchenkov , a. matsko , d. strekalov , v. ilchenko , and l. maleki , `` mode filtering in optical whispering gallery resonators , '' electron . lett . * 41 * , 495 ( 2005 ) . g. lin , b. qian , f. orucevi , y. candela , j .- b . jager , z. cai , v. lefvre - seguin , and j. hare , `` excitation mapping of whispering gallery modes in silica microcavities . '' opt . lett . * 35 * , 583585 ( 2010 ) . d. keng , x. tan , and s. arnold , `` whispering gallery micro - global positioning system for nanoparticle sizing in real time , '' appl phys . lett . * 105 * , 071105 ( 2014 ) . m. l. gorodetsky and v. s. ilchenko , `` high - q optical whispering - gallery microresonators : precession approach for spherical mode analysis and emission patterns with prism couplers , '' opt . commun . * 113 * , 133143 ( 1994 ) . a. savchenkov , a. matsko , v. s. ilchenko , d. strekalov , and l. maleki , `` direct observation of stopped light in a whispering - gallery - mode microresonator , '' phys . rev . a * 76 * , 023816 ( 2007 ) . t. carmon , h. g. l. schwefel , l. yang , m. oxborrow , a. d. stone , and k. j. vahala , `` static envelope patterns in composite resonances generated by level crossing in optical toroidal microcavities , '' phys . rev . lett . * 100 * , 103905 ( 2008 ) . c. dong , y. xiao , y. yang , z. han , g. guo , and l. yang , `` directly mapping whispering gallery modes in a microsphere through modal coupling and directional emission , '' chin . opt . lett . * 6 * , 300302 ( 2008 ) . f. sedlmeir , m. hauer , j. frst , g. leuchs , and h. g. l. schwefel , `` experimental characterization of an uniaxial angle cut whispering gallery mode resonator , '' opt . express * 21 * , 2394223949 ( 2013 ) . s. schiller and r. l. byer , `` high - resolution spectroscopy of whispering gallery modes in large dielectric spheres , '' opt . lett . * 16 * , 11381140 ( 1991 ) . s. preu , h. g. l. schwefel , s. malzer , g. h. dhler , l. j. wang , m. hanson , j. d. zimmerman , and a. c. gossard , `` coupled whispering gallery mode resonators in the terahertz frequency range , '' opt . express * 16 * , 73367343 ( 2008 ) . j. li , h. lee , k. yang , and k. vahala , `` sideband spectroscopy and dispersion measurement in microcavities , '' opt . express * 20 * , 2633726344 ( 2012 ) . a. n. oraevsky , `` whispering - gallery waves , '' quant . electron . * 32 * , 377400 ( 2002 ) . m. l. gorodetsky and y. a. demchenko , `` accurate analytical estimates of eigenfrequencies and dispersion in whispering - gallery spheroidal resonators , '' proc . spie * 8236 * , 823623 ( 2012 ) . i. breunig , b. sturman , f. sedlmeir , h. g. l. schwefel , and k. buse , `` whispering gallery modes at the rim of an axisymmetric optical resonator : analytical versus numerical description and comparison with experiment , '' opt . express * 21 * , 3068330692 ( 2013 ) . m. l. gorodetsky and v. s. ilchenko , `` optical microsphere resonators : optimal coupling to high - q whispering - gallery modes , '' j. opt . soc . am . b * 16 * , 147154 ( 1999 ) . m. gorodetsky and a. fomin , `` geometrical theory of whispering - gallery modes , '' ieee j. sel . topics quantum electron . * 12 * , 3339 ( 2006 ) . u. schlarb and k. betzler , `` influence of the defect structure on the refractive indices of undoped and mg - doped lithium niobate , '' phys . rev . b * 50 * , 751757 ( 1994 ) .
in the study of high - temperature superconductivity in the cuprates , there has been intense interest in the interplay between antiferromagnetism and superconductivity . the undoped parent compound , which is a mott insulator , is an antiferromagnetic ( af ) long - range ordered state . upon carrier doping , the af state is converted into the high - temperature superconducting ( sc ) state . in the mechanism of superconductivity , the af correlation is believed to play an important role . in order to investigate the af correlation effect on superconductivity , multilayer systems are useful . in multilayer systems , the number of cuo@xmath0 layers per unit cell , @xmath1 , takes @xmath2 . from systematic studies of the @xmath1 dependence of sc transition temperature @xmath3 , it is found that the maximum @xmath3 is obtained @xcite for the case of @xmath4 . since coupling between the cuo@xmath0 planes within the unit cell is stronger than that between the cuo@xmath0 planes in different unit cells , the af correlation can be strong in multilayer systems . there is also charge imbalance over the cuo@xmath0 layers as revealed by the @xmath5cu knight shifts , and thus the multilayer systems form a natural heterostructure at an intermediate carrier - concentration region.@xcite from nuclear magnetic resonance ( nmr ) experiments it was suggested that af moments survived even in a sc phase.@xcite a possibility of a phase separation is ruled out because the nmr signal associated with the paramagnetic state was not observed.@xcite an important issue here is whether sc order coexists with af order @xcite or not . recently , shimizu _ et al_. studied the temperature dependence of the af moment in a five - layered cuprate ba@xmath0ca@xmath6cu@xmath7o@xmath8(f , o)@xmath0 by nmr measurements.@xcite they reported that the af moment in the outer layer is enhanced below @xmath3 . such an enhancement of the af moment below @xmath3 does not appear in the coexistence phase of other superconductors such as iron arsenide superconductor ba@xmath0(fe@xmath9co@xmath10)@xmath0as@xmath0 , where the af moment reduces at @xmath3.@xcite therefore , the enhancement seen in the five - layered cuprate is not necessarily common to all of the coexistence phase of superconductors . noting that the iron arsenide is a single - layer system , we may speculate that the multilayer nature is essential for the enhancement , and thus need to clarify the mechanism of the af moment enhancement triggered by the sc transition under the presence of multilayers . in this paper , we study the interplay between sc and af orders within a mean field theory . to describe a multilayer system , where there are two types of layers associated with outer planes and inner planes , we consider a bilayer system with different electronic correlations in each layer . by controlling interaction parameters for the af correlation and the sc correlation , we study the proximity effect between two layers with different order and also examine the possibility of coexistence of sc and af orders . we find that the proximity effect leads to an enhancement of af order below the sc transition temperature that is consistent with the experiment.@xcite by contrast , we find qualitatively different behaviors of the order parameters for the coexistence case . this paper is organized as follows . in sec.[sec_model ] , we introduce the bilayer hamiltonian and describe the mean field theory . in our model , a coexistence phase of sc and af orders is possible within a single layer . this coexistence phase is described in sec.[sec_single ] . in sec.[sec_bilayer ] , we describe the results about the proximity effect in the bilayer system . we show that the temperature dependence of the af moment is consistent with the experiment.@xcite we also examine coexistence phases and show that the temperature dependence of the order parameters is quite different from the experiment . finally , sec.[sec_summary ] is devoted to summary and discussions . we consider a bilayer system with interactions which stabilize sc and/or af order . the two layers are coupled through an interlayer tunneling . the hamiltonian is given by @xmath11 where the interlayer tunneling term @xmath12 is @xmath13 here @xmath14 ( @xmath15 ) creates ( annihilates ) electrons with in - plane momentum @xmath16 and spin @xmath17 at layer @xmath18 . we assume that the interlayer hopping matrix @xmath19 is independent of @xmath16 . the hamiltonian for the @xmath18 layer is @xmath20 with @xmath21 . ( hereafter we set the lattice constant to unity . ) the hopping of electrons within each layer is restricted to the nearest neighbors given by @xmath22 . energies are measured in units of @xmath22 in the following analysis . we focus on the half - filing case in each layer so that we set the chemical potential @xmath23 . this means that we neglect charge redistribution between the two layers . the second term of @xmath24 describes the af interaction . the operator @xmath25 ( @xmath26 ) creates ( annihilates ) electrons at site @xmath27 . the third term of @xmath24 with @xmath28 describes the interaction for @xmath29-wave pairing . we define the @xmath30-wave sc order parameter in the @xmath18 layer with @xmath31 sites by @xmath32 with @xmath33 the af order parameter in the @xmath18 layer is defined by @xmath34 where the summation with respect to @xmath16 is taken over the reduced brillouin zone @xmath35 and the nesting vector is @xmath36 . by using the order parameters , the mean field hamiltonian at the @xmath18 layer reads @xmath37 where @xmath38 and @xmath39 based on the mean field hamiltonian of the whole system , @xmath40 , we solve the mean - field eqs . ( [ eq_delta ] ) and ( [ eq_m ] ) numerically using the @xmath41 discretized brillouin zone . before going into the analysis of the bilayer system , we examine a coexisting phase in a single layer system described by @xmath42 . reflecting the difference in symmetry between the af gap created by @xmath43 and the sc gap @xmath44 , the coexistence phase of sc and af orders can be stabilized.@xcite the situation is similar to the slave - boson mean field theory of the @xmath22-@xmath45 model @xcite and the hubbard model in the strong - coupling limit.@xcite figure [ fig_single](a ) shows the parameter range of @xmath46 for the coexistence phase at @xmath47 . for @xmath48 the system is a pure af state while for @xmath49 the system is a pure sc state.@xcite so the coexistence phase appears for @xmath50 . for the case of @xmath51 , the parameter for the coexistence changes as @xmath52 . in order to confirm that the state with @xmath53 and @xmath54 is the global minimum of the free energy , we computed the following energy at @xmath55 for @xmath56 and @xmath57 : @xmath58 where @xmath59 runs over the all eigenstates of the mean field hamiltonian and @xmath60 are the eigenenergies . we examined several cases and confirmed that the coexistence phase solution corresponds to the global minimum of the energy . we examined the @xmath61-wave case as well , but there is no coexistence phase . there are two types of coexistence phases . one is the phase with strong af order and weak sc order , resulting in the af transition temperature @xmath62 , and the other is the phase with @xmath63 . figure [ fig_single](b ) shows the temperature dependence of the order parameters in the coexistence phase with @xmath62 at @xmath64 and @xmath65 . for comparison , the pure af case at @xmath66 and @xmath65 and the pure sc case at @xmath64 and @xmath67 are also shown . for this choice of the parameters , @xmath68 is slightly higher than @xmath3 . therefore , the system first exhibits af order upon decreasing temperature . @xmath3 is somewhat reduced because of the presence of af order . the occurrence of the sc order also affects the af order : the temperature dependence of @xmath69 in fig . [ fig_single](b ) deviates from the pure af case at @xmath3 , resulting in the reduction of @xmath69 . this behavior is in contrast to the case of the multilayer cuprates where the enhancement of the af order is observed . @xcite we note that , in the coexistence phase with @xmath70 , @xmath3 is the same as the value of the pure sc case but @xmath68 is reduced from the pure af case value . in this section , we study the bilayer system . our purpose is twofold . first , we study the proximity effect in the bilayer system . second , we examine the stability of the coexistence phase in a single layer under the presence of inter - layer tunneling . to start with , we examine the interlayer tunneling effect . depending on the value of @xmath19 , there are a strong @xmath19 regime and a weak @xmath19 regime . figure [ fig_tp ] shows the @xmath19 dependence of the af order parameters @xmath71 and @xmath72 at @xmath73 and @xmath74 with @xmath75 . for @xmath76 , we see that the values of the order parameters are not so much affected by the increase of @xmath19 . this weak @xmath19 regime is not suitable for describing multilayer systems because each layer is almost independent . in fact , the change of @xmath71 and @xmath72 in the weak @xmath19 regime is described by the second - order perturbation theory with respect to @xmath19 . at @xmath77 , there is a first order transition between the weak @xmath19 regime and the strong @xmath19 regime as shown in fig . [ fig_tp ] . for @xmath78 , the order parameters exhibit strong @xmath19 dependence . in this strong @xmath19 regime , @xmath19 is larger than the excitation gap created by af order . therefore , the order parameters are reduced due to the change of the fermi - surface topology . in the large @xmath19 limit , the noninteracting single - body electron states are well described by the bonding state and the anti - bonding state . the fermi surface splits into two pockets centered at the @xmath79 point and @xmath80 point . qualitatively similar behaviors are found in the @xmath19 dependence of the sc order parameters . in the following analysis , we focus on this strong @xmath19 regime and set @xmath81 . now we investigate the bilayer system . we consider three cases . in all cases , we assume @xmath82 and @xmath83 for @xmath84 . therefore , the intrinsic order in the @xmath84 layer is restricted to af order . for the @xmath85 layer , we assume @xmath86 and @xmath87 . the three cases are ( i ) @xmath88 , ( ii ) @xmath89 , and ( iii ) @xmath90 . it turns out that the ground state of case ( i ) consists of sc order in the @xmath85 layer and af order in the @xmath84 layer when @xmath91 . the ground states of cases ( ii ) and ( iii ) are the coexistence state of sc and af orders in the @xmath85 layer and only af order in the @xmath84 layer . cases ( ii ) and ( iii ) are distinguished by the strength of the af order in the two layers : @xmath92 in case ( ii ) , while @xmath93 in case ( iii ) . figure [ fig_case_i](a ) is a schematic view of case ( i ) . the temperature dependence of the order parameters is shown in figs . [ fig_case_i](b)-(d ) for different values of @xmath94 . for @xmath95 , there is no sc order in the @xmath85 layer . intrinsic af order @xmath72 appears in the @xmath84 layer for @xmath96 as shown in fig . [ fig_case_i](d ) . as a result of the proximity effect , a finite value of @xmath71 is induced as shown in fig . [ fig_case_i](c ) . this induced @xmath71 decreases with increasing @xmath94 , because the intrinsic sc order by @xmath94 competes with the induced af order . on the other hand , the intrinsic af order @xmath72 in the @xmath84 layer increases with increasing @xmath94 . at @xmath97 there is a sc transition in the @xmath85 layer as shown in fig . [ fig_case_i](b ) . in the presence of the non - zero sc order parameter , @xmath98 , in the @xmath85 layer , a finite value of @xmath99 is induced in the @xmath84 layer ( not shown ) because of the proximity effect . the sc transition affects @xmath72 . as shown in fig . [ fig_case_i](d ) , there is a clear enhancement of @xmath72 below the sc transition temperature @xmath100 for the case of @xmath101 . a similar enhancement is also found in the case of @xmath102 below @xmath103 . this enhancement of @xmath72 below @xmath3 is consistent with the experiment in the multilayer cuprate.@xcite for @xmath104 this enhancement is masked because @xmath3 is higher than @xmath68 . the enhancement of af order below @xmath3 is also found in the case ( ii ) ( fig . [ fig_case_ii](a ) ) . figures [ fig_case_ii](b)-(d ) show the temperature dependence of the order parameters for different values of @xmath94 . for the case of @xmath105 , there is no intrinsic sc order in the @xmath85 layer . for the case of @xmath106 , there is a sc transition at @xmath107 . below @xmath3 , @xmath72 is clearly enhanced , although the enhancement is much reduced compared to the case ( i ) . similar behaviors are observed for @xmath108 . again this behavior is consistent with the experiment.@xcite now we examine the case ( iii ) ( fig . [ fig_case_iii](a ) ) . the temperature dependence of the order parameters is shown in figs . [ fig_case_iii](b)-(d ) . in this case , we observe quite different behaviors of the order parameters compared to the cases ( i ) and ( ii ) . in particular , the superconducting transition temperature @xmath109 is always larger than the af transition temperature @xmath68 . there is no coexistence phase when @xmath110 . for @xmath111 , @xmath98 increases as @xmath112 decreases as shown in fig . [ fig_case_iii](b ) . below @xmath68 , @xmath98 is suppressed . furthermore , the coexistence phase is limited to a finite range of temperature for @xmath113 . meanwhile the order parameters @xmath71 and @xmath72 increase monotonically as the temperature decreases as shown in figs . [ fig_case_iii](c ) and ( d ) . these temperature dependences are qualitatively different from the experimentally observed one . from this observation one may conclude that it is unlikely that there is a coexistence phase of intrinsic sc and af orders in a layer among coupled multilayers . although the coexistence phase of af and sc orders appears in all cases ( i)-(iii ) , the origin of af order in the sc layer is different . what makes the difference between the case ( iii ) and the case ( ii ) is that in the case ( iii ) af order in the @xmath85 layer with sc is intrinsic order but not induced by the other layer . meanwhile in the case ( ii ) af order in the @xmath85 layer with sc is induced order by af order in the @xmath84 layer due to the proximity effect . so far we have studied the system with the electron hopping restricted to the nearest neighbors . in order to examine the system with a realistic fermi surface,@xcite we consider the model with @xmath114 here we choose @xmath115 and @xmath116 . we take the chemical potential as @xmath117 , which corresponds to @xmath118 doping in the normal and non - magnetic state . the result for the case ( i ) above with @xmath88 and @xmath119 is shown in fig . [ fig_case_i_fs ] . the temperature dependence of the order parameters is shown in figs . [ fig_case_i_fs](b)-(d ) for different values of @xmath94 . we observe qualitatively similar behaviors of the order parameters to the case ( i ) shown in fig . [ fig_case_i ] . there is a clear enhancement of @xmath72 below the sc transition . intrinsic af order @xmath72 appears in the @xmath84 layer for @xmath120 as shown in fig . [ fig_case_i_fs](d ) . this intrinsic order is enhanced below the sc transition . for example , there is the sc transition at @xmath121 for @xmath122 as shown in fig . [ fig_case_i_fs](b ) . for @xmath123 , @xmath72 is enhanced compared with the @xmath124 case . meanwhile , @xmath71 is suppressed as shown in fig . [ fig_case_i_fs](c ) . the exceptional case is @xmath97 . the temperature dependence of @xmath72 is similar to the other cases but the temperature dependence of @xmath71 is qualitatively different . the value of @xmath71 is enhanced below the sc transition . this discrepancy is probably associated with the change of the fermi surface shape . to summarize , we have studied the proximity effect and the possibility of coexistence of af and sc orders in a bilayer system . our mean field theory suggests that the experimentally observed enhancement of af order below @xmath3 @xcite is associated with the proximity effect . in contrast , if we assume a coexistence phase in a layer among coupled multilayers , the temperature dependence of the order parameters is qualitatively different from the experimentally observed one . we believe that this result is not so much affected by the shape of the fermi surface . as we have shown in fig . [ fig_case_i_fs ] , the result for a realistic fermi surface with finite doping is qualitatively the same as that for the half - filling case . so we expect qualitatively the same proximity effect as long as we neglect the possibility of stabilizing other orders , such as a charge - density wave or the so - called @xmath125-triplet pairing.@xcite the absence of the @xmath125-triplet pairing is the unique property of the half - filling case with @xmath126 . however , there is no experimental evidence for the @xmath125-triplet pairing to the best of our knowledge . this work was supported by the grant - in - aid for scientific research from the ministry of education , culture , sports , science , and technology of japan ; the global center of excellence ( coe ) program the next generation of physics , spun from university and emergence " ; and the yukawa institutional program for quark - hadron science at yukawa institute of theoretical physics . in fig . [ fig_single](a ) @xmath69 has tiny values for @xmath49 but this is a finite size effect . by increasing the number of brillouin zone points taken in the numerical calculation , the finite values of @xmath69 for @xmath49 are suppressed . r. m. fernandes , d. k. pratt , w. tian , j. zarestky , a. kreyssig , s. nandi , m. g. kim , a. thaler , n. ni , p. c. canfield , r. j. mcqueeney , j. schmalian , and a. i. goldman , phys . b * 81 * , 140501(r ) ( 2010 ) . a. damascelli , z. hussain , and z .- x . shen , rev . phys . * 75 * , 473 ( 2003 ) . a. a. kordyuk and s. v. borisenko , low temp . phys . * 32 * , 298 ( 2006 ) . t. tohyama and s. maekawa , supercond . * 13 * , r17 ( 2000 ) .	motivated by the recent experiment in multilayered cuprate superconductors reporting the enhancement of antiferromagnetic order below the superconducting transition temperature , we study the proximity effect of the antiferromagnetic correlation in a bilayer system and also examine the possibility of a coexistence of antiferromagnetic order and superconductivity . we present the result of mean field theory that is consistent with the experiment and supports the proximity effect picture .
in recent years several short - baseline neutrino oscillation experiments have found anomalies which may require an extension of the standard three - neutrino mixing framework which describes the neutrino oscillations observed in solar , atmospheric and long - baseline experiments ( see @xcite ) . in this paper we consider the gallium anomaly @xcite and the reactor anomaly @xcite , which indicate that electron neutrino and antineutrinos may disappear at short distances anomalies will be discussed elsewhere @xcite . ] . such disappearance may be explained by the presence of at least one massive neutrino at the ev scale , which drives short - baseline neutrino oscillations generated by a squared - mass difference which is much larger than the squared - mass difference operating in the solar , atmospheric and long - baseline neutrino oscillation experiments . we consider 3 + 1 neutrino mixing , which is the minimal extension of three - neutrino mixing which can explain the gallium and reactor anomalies . since from the lep measurement of the invisible width of the @xmath5 boson @xcite we know that there are only three light active flavor neutrinos the additional neutrino in the 3 + 1 framework is sterile . in this paper we discuss the implications of the recent @xmath3 measurement in ref . @xcite , which give information on the neutrino cross section in gallium experiments . we take also into account the most updated calculation of the reactor neutrino fluxes presented in the recent white paper on light sterile neutrinos @xcite . we present also a detailed discussion of the connection between the results of the fit of neutrino oscillation data and the results of @xmath4-decay experiments ( considering the mainz data presented very recently in ref . @xcite ) and neutrinoless double-@xmath4 decay experiments ( considering the recent exo bound in ref . @xcite and the controversial positive result in ref . @xcite ) . we consider 3 + 1 neutrino mixing as an extension of standard three - neutrino mixing . the mixing of the three active flavor neutrino fields @xmath0 , @xmath6 , @xmath7 and one sterile neutrino field @xmath8 is given by @xmath9 where @xmath10 is the unitary @xmath11 mixing matrix ( @xmath12 ) and each of the four @xmath13 s is a massive neutrino field with mass @xmath14 . we consider the squared - mass hierarchy @xmath15 with @xmath16 , such that @xmath17 generates the very - long - baseline oscillations observed in solar neutrino experiments and in the kamland reactor antineutrino experiment , @xmath18 generates the long - baseline oscillations observed in atmospheric neutrino experiments and in long - baseline accelerator and reactor neutrino and antineutrino experiments , and @xmath19 generates short - baseline oscillations . the effective survival probability at a distance @xmath20 of electron neutrinos and antineutrinos with energy @xmath21 in short - baseline neutrino oscillation experiments is given by ( see refs . @xcite ) @xmath22 with the transition amplitude @xmath23 the plan of the paper is as follows . in section [ gallium anomaly ] we discuss in detail the gallium @xmath0 anomaly @xcite and the implications of the important recent @xmath3 measurement in ref . @xcite . in section [ fit of gallium and reactor data ] we present the results of the combined analysis of gallium data with reactor @xmath1 data , taking into account the reactor @xmath1 anomaly @xcite . in section [ solar neutrino constrait ] we discuss the solar neutrino constraint on short - baseline @xmath0 disappearance @xcite . in section [ global fit ] we present the results of the global fit of @xmath0 and @xmath1 disappearance data , which includes also @xmath24 scattering data @xcite . we confront these results with the bounds on the heavy neutrino mass given by the data of @xmath4-decay experiments @xcite and neutrinoless double-@xmath4 decay experiments @xcite . finally , in section [ conclusions ] we draw our conclusions . . [ tab : src ] energy ( @xmath21 ) and branching ratio ( @xmath25 ) of the neutrino lines produced in the electron - capture decay of @xmath26 and @xmath27 . [ cols="<,^,^,^,^,^,^ " , ] the results of the combined fit of gallium and reactor data are presented in tab . [ tab : fit - rea ] and figs . [ fig : fit - ghk - rea][fig : fit - ghf - rea ] . from these figures , one can see that the allowed regions in the @xmath28@xmath19 in the three cases that we have considered for the fit of gallium data are quite similar , and the best - fit values of @xmath28 and @xmath19 are equal ( see tab . [ tab : fit - rea ] ) . this is due to a dominance of reactor data , which are more numerous and have smaller uncertainties . hence , in the following we consider only the ff case , which is the one which is more compatible with reactor data , because it agrees more than the hk and hf cases with the reactor exclusion of large values of @xmath28 . this better agreement is quantified by the larger parameter goodness - of - fit @xcite in tab . [ tab : fit - rea ] . in this section we discuss the upper bound for @xmath28 which can be obtained from the data of solar neutrino experiments @xcite and from the data of the kamland very - long - baseline reactor antineutrino experiment @xcite , which are sensitive to oscillations generated by the small squared - mass difference @xmath29 . since the event rates measured in these experiments are well described by standard three - neutrino mixing , the data allow us to constrain the corrections due to active - sterile neutrino mixing , which affects the electron neutrino and antineutrino survival probability and generates transitions into sterile neutrinos @xcite . as explained in the following , the almost degenerate effects in the solar and kamland experiments @xcite of @xmath30 , which determines @xmath28 through eq . ( [ survisin ] ) , and @xmath31 can be resolved by using the recent determination of @xmath31 in the daya bay @xcite and reno @xcite long - baseline reactor antineutrino experiments . the effective survival probability of electron neutrinos and antineutrinos in the solar and kamland experiments is given by @xcite @xmath32 where @xmath33 is the two - neutrino survival probability generated by @xmath29 . considering small values of @xmath31 and @xmath30 , we have @xmath34 in vacuum the two - neutrino survival probability @xmath33 has the standard two - neutrino form which does not depend on @xmath31 and @xmath30 . therefore , in the kamland experiment @xmath31 and @xmath30 have the same effect of suppressing the electron neutrino and antineutrino survival probability . for solar neutrinos , the main effect of @xmath31 and @xmath30 is the same as in the vacuum case , but there are corrections caused by the modifications of @xmath33 due to the decrease of @xmath35 if @xmath36 and/or @xmath37 and to the contribution of the neutral current potential @xmath38 which is not felt by the sterile neutrino during propagation in matter . in order to describe this effect , we neglect possible cp - violating phases in the mixing matrix and we parameterize it as ( see also @xcite ) @xmath39 where @xmath40 is the real orthogonal matrix ( @xmath41 ) which operates a rotation in the @xmath42-@xmath43 plane by an angle @xmath44 : @xmath45_{rs } = \delta_{rs } \null & \null + \left ( c_{ab } - 1 \right ) \left ( \delta_{ra } \ , \delta_{sa } + \delta_{rb } \ , \delta_{sb } \right ) \nonumber \\ \null & \null + s_{ab } \left ( \delta_{ra } \ , \delta_{sb } - \delta_{rb } \ , \delta_{sa } \right ) \ , , \label{rab}\end{aligned}\ ] ] with @xmath46 and @xmath47 . in this parameterization the electron line of the mixing matrix is given by @xmath48 hence , this is an extension of the standard three - neutrino mixing parameterization of the electron line ( see @xcite ) with the addition of @xmath0-@xmath49 mixing parameterized by @xmath50 . this parameterization is also convenient because @xmath51 since the sterile line of the mixing matrix is more complicated , it is convenient to write its first two elements as @xmath52 with @xmath53 the adiabatic two - neutrino survival probability @xmath33 is given by @xcite @xmath54 where @xmath55 is the effective mixing angle at neutrino production , which is given by @xmath56 the mixing angle @xmath57 between the vacuum mass basis and the effective mass basis in matter is given by @xmath58 here @xmath59 is the matter potential given by @xmath60 where @xmath61 and @xmath38 are the standard charged - current and neutral - current matter potentials . the angle @xmath62 is given by @xmath63 therefore , for @xmath64 the contributions of @xmath65 and @xmath66 to the matter effects are almost degenerate . there is only a small difference of their contributions due to @xmath5 in eq . ( [ zste ] ) . in solar neutrino measurements the degeneracy of the effects of @xmath31 and @xmath30 is also slightly broken by the sno neutral current measurement , which is sensitive to the total probability of @xmath0 transitions into active neutrinos , which by unitarity is given by @xmath67 , with @xmath68 here , the adiabatic two - neutrino transition probability @xmath69 is given by one can check that in the limit of two - neutrino @xmath0-@xmath8 mixing the unitarity relation @xmath70 is satisfied . in this case , @xmath71 and @xmath72 . ] @xcite @xmath73 the degeneracy of @xmath31 and @xmath30 is broken by their different effects in @xmath74 , @xmath75 and in the last term of eq . ( [ pes - sun ] ) . luckily , the recent determination of the value of @xmath76 in the daya bay @xcite and reno @xcite experiment allow us to obtain information on the value of @xmath77 without much uncertainty due to @xmath76 . in the 3 + 1 mixing scheme under consideration the effective long - baseline @xmath1 survival probability in the daya bay and reno far detectors is given by @xcite @xmath78 since the oscillations due to @xmath79 are averaged and the oscillations due to @xmath80 are negligibly small . this survival probability depends on @xmath50 , but the daya bay and reno collaborations measured the ratio of the probability ( [ peelblfar ] ) in the far detectors and the probability ( [ peelblfar ] ) with @xmath81 in the near detectors , @xmath82 since the contribution of small values of @xmath50 to the measured ratio is of order @xmath83 @xcite , in practice the value of @xmath76 determined by the daya bay and reno collaborations with a three - neutrino mixing survival probability is accurate also in the 3 + 1 scheme under consideration . nevertheless , since we have the possibility , in our analysis we took into account the exact ratio of the far and near survival probabilities ( [ peelblfar ] ) and ( [ peelblnear ] ) by including the least - squares function @xmath84 of the far / near relative measurements of daya bay and reno in the total solar and reactor least - squares function @xmath85 for the calculation of the solar least - squares function @xmath86 , we considered the radiochemical @xmath87 @xcite and @xmath88 @xcite experiments , the day and night energy spectra of all four phases of super - kamiokande @xcite , the day and night energy spectra of the sno @xmath89 phase @xcite , the charged - current and neutral - current rates of the sno salt @xcite and ncd @xcite phases , and the rates of low energy @xmath90 @xcite and pep @xcite solar neutrinos from the borexino experiment . we did not use the sno data obtained with the low energy threshold analysis @xcite and those obtained with the combined analysis @xcite because both analyses assumed a three - parameter polynomial survival probability which is not appropriate for the sterile neutrino analysis . the solar neutrino fluxes are taken from the bp2004 @xcite standard solar model , except for the normalization of the solar @xmath91 neutrino flux , which is considered as a free parameter determined by the minimization of @xmath92 ( as usual , because of its large theoretical uncertainties ) . the kamland least - squares function @xmath93 has been calculated using the energy spectrum reported in @xcite with a total exposure of @xmath94 target - proton - year . in our analysis , we used the @xmath0 survival probability ( [ peesun1 ] ) and the @xmath95 transition probability ( [ pes - sun ] ) taking into account as parameters the squared - mass difference @xmath17 and the five relevant mixing angles @xmath96 , @xmath76 , @xmath50 , @xmath97 , @xmath98 ( solar and kamland oscillations are independent from @xmath99 , because @xmath6 and @xmath7 are indistinguishable ; see @xcite ) . the six - dimensional parameter space is explored with a markov chain monte carlo sampling in order to minimize the total @xmath92 in eq . ( [ chi2sun ] ) . marginalized @xmath100 as a function of @xmath28 obtained from the fit of solar and kamland data with and without daya bay and reno data . ] since the best fit is obtained for @xmath101 , the marginalized @xmath100 shown in fig . [ fig : sun ] give stringent constraints on the value of @xmath102 . in fig . [ fig : sun ] we have plotted the @xmath103 obtained with and without including the daya bay and reno data . one can see that these data are useful in order to tighten the upper bound on @xmath28 . in this section we present the results of the global fit of electron neutrino and antineutrino disappearance data , which includes the gallium and reactor data discussed respectively in sections [ gallium anomaly ] and [ fit of gallium and reactor data ] , the solar neutrino constraint discussed in section [ solar neutrino constrait ] , and the karmen @xcite and lsnd @xcite @xmath24 scattering data @xcite , with the method discussed in ref . @xcite . allowed 95% cl regions in the @xmath28@xmath19 plane obtained from the separate fits of gallium , reactor , solar and @xmath2 scattering data and from the combined fit of all data . the best - fit points corresponding to @xmath104 are indicated by crosses . ] allowed regions in the @xmath28@xmath19 plane and marginal @xmath105 s for @xmath28 and @xmath19 obtained from the global fit of @xmath0 and @xmath1 data . the best - fit point corresponding to @xmath104 is indicated by a cross . ] figure [ fig : fit - glo - sup ] shows a comparison of the allowed 95% cl regions in the @xmath28@xmath19 plane obtained from the separate fits of gallium , reactor , solar and @xmath2 scattering data and from the combined fit of all data . one can see that the separate allowed regions overlap in a band delimited by @xmath106 and @xmath107 , which is included in the globally allowed 95% cl region . figure [ fig : fit - glo ] shows the globally allowed regions in the @xmath28@xmath19 plane and the marginal @xmath105 s for the two oscillation parameters . the best - fit point is at a relatively large value of @xmath19 , @xmath108 with @xmath109 , corresponding to a @xmath110% goodness - of - fit . however , there is a region allowed at @xmath111 around @xmath112 and @xmath113 . the slight preference of the global fit for @xmath114 with respect to @xmath112 ( see the marginal @xmath105 for @xmath19 in fig . [ fig : fit - glo ] ) , which is preferred by gallium and reactor data ( see tabs . [ tab : fit - gal ] and [ tab : fit - rea ] and figs . [ fig : fit - ghk][fig : fit - ghf ] , [ fig : fit - rea ] and [ fig : fit - ghk - rea][fig : fit - ghf - rea ] ) , is due to the @xmath2 scattering data , which prefer larger values of @xmath19 ( see the discussion in ref . @xcite ) . comparing the minimum of the @xmath92 of the global fit with the sum of the minima of the @xmath92 of the separate fits of gallium , reactor , solar and @xmath2 scattering data , we obtained @xmath115 , with @xmath116 degrees of freedom , which gives a parameter goodness - of - fit of 4% . therefore , the compatibility of the four data sets is acceptable . relative deviation of the kurie plot in @xmath4-decay for some points in the allowed regions of fig . [ fig : fit - glo ] . ] the results of the global fit , as well as the results of the fits of gallium and reactor data , lead to lower limits for @xmath117 , but there is no upper limit for @xmath117 in figs . [ fig : fit - ghk][fig : fit - ghf ] , [ fig : fit - rea ] , [ fig : fit - ghk - rea][fig : fit - ghf - rea ] and [ fig : fit - glo ] . hence , one can ask if there are other measurements which constrain large values of @xmath117 . the answer is positive and comes from the measurements of the effects of heavy neutrino masses on electron spectrum in @xmath4-decay far from the end - point , from the results of neutrinoless double-@xmath4 decay experiments for majorana neutrinos , and from cosmological measurements . in this discussion we consider the mass hierarchy @xmath118 which implies @xmath119 let us consider first @xmath4-decay experiments . the ratio of the kurie function @xmath120 in @xmath4-decay for the case of a heavy neutrino @xmath49 and that corresponding to massless neutrinos is given by @xcite @xmath121 where @xmath122 is the kinetic energy of the electron , @xmath123 is the @xmath124-value of the decay , @xmath125 is the heaviside step function , and we have neglected the contribution of the three light neutrinos @xmath126 , @xmath127 , @xmath128 . figure [ fig : kur ] shows the relative deviation of the kurie plot with respect to the massless case for some points in the allowed regions of fig . [ fig : fit - glo ] . one can see that in order to see the effect of @xmath129 , beta - decay experiments must have a sensitivity to the relative deviation of the kurie plot of the order of a percent or better for @xmath130 . in 2001 the genoa @xmath131 beta - decay experiment @xcite searched for deviation of the electron spectrum due to a heavy neutrino with a mass from 50 to 1000 ev . from fig . 3 of ref . @xcite one can see that the 95% cl upper bound for @xmath129 is about 300 ev if @xmath132 , which implies a very large upper limit on @xmath117 of about @xmath133 . comparison of the mainz @xmath4-decay bound ( curves in the top - right part of the figure ) with the allowed regions in the @xmath28@xmath19 plane obtained from the global fit of @xmath0 disappearance data ( same as in fig . [ fig : fit - glo ] ) . ] allowed regions in the @xmath28@xmath19 plane obtained from the combined fit of @xmath0 disappearance and mainz @xmath4-decay data . ] very recently , the mainz collaboration released new data obtained with the phase ii of the mainz neutrino mass experiment @xcite which constrain the value of @xmath134 for @xmath135 between about 10 and @xmath136 . figure [ fig : con - mnz ] shows the constraints in the @xmath28@xmath19 plane that we obtained with a @xmath92 analysis of the mainz data in @xcite . from the comparison with the allowed regions obtained from the global fit of @xmath0 disappearance data shown in fig . [ fig : con - mnz ] one can see that the mainz data constrain @xmath19 to be smaller than about @xmath137 at about 90% cl . this is confirmed by the results of the combined fit shown in fig . [ fig : con - osc - mnz ] . the katrin experiment ( see @xcite ) , which will start in 2015 @xcite , may be able to improve dramatically the upper limits on @xmath129 and maybe see its effects on the electron spectrum @xcite . marginal @xmath100 as a function of @xmath138 obtained from the global fit . the vertical green band represents the currently most stringent 90% cl upper bound for @xmath138 in the no - cancellation case ( [ no - cancellation ] ) obtained from the 90% cl exo bound on @xmath139 taking into account nuclear matrix element uncertainties @xcite . the vertical dark - green band corresponds to the @xmath111 klapdor - kleingrothaus et al . range of @xmath139 @xcite . ] the heavy neutrino mass @xmath129 has also an effect in neutrinoless double-@xmath4 decay ( see @xcite ) , if massive neutrinos are majorana particles ( see @xcite ) . considering eq . ( [ m4 ] ) , the contribution of the heavy neutrino mass @xmath129 to the effective majorana mass @xmath140 is given by @xmath141 figure [ fig : mbb - chi ] shows the marginal @xmath100 as a function of @xmath138 obtained from the global fit . one can see that @xmath138 is bounded from below and it is likely to be larger than about @xmath142 , a value which may be reached in the next generation of neutrinoless double-@xmath4 decay experiment ( see @xcite ) . of course , if the three light neutrinos @xmath126 , @xmath127 , @xmath128 are quasi - degenerate at a mass scale larger than about @xmath142 the contribution of @xmath138 can cancel with that of the three light neutrinos . such an unfortunate cancellation can also happen if the masses of the three light neutrinos follow an inverted hierarchy @xcite , since in that case their contribution to the effective majorana mass is @xcite @xmath143 on the other hand , no cancellation is possible in the case of a normal hierarchy , for which @xcite @xmath144 and the contribution of @xmath138 is dominant if it is larger than about @xmath142 . figure [ fig : mbb ] shows the allowed values of @xmath139 as functions of the lightest neutrino mass in the two 3 + 1 schemes with a normal ( left ) and inverted ( right ) mass spectra of the three light neutrinos . we have drawn also the curves which delimit the three - neutrino mixing allowed regions @xcite . one can see that practically the situation is reversed with respect to the three - neutrino mixing case ( see also the discussion in @xcite ) , in which @xmath139 is predicted to be large in the inverted spectrum and can vanish in the normal spectrum . in the 3 + 1 case @xmath139 can have any value in the inverted spectrum , whereas in the normal spectrum it is likely to be large if the three light neutrino masses are hierarchical , i.e. if @xmath145 . comparison of the allowed regions in the @xmath28@xmath19 plane obtained from the global fit of @xmath0 disappearance data ( same as in fig . [ fig : fit - glo ] ) and the bound in the no - cancellation case ( [ no - cancellation ] ) corresponding to the 90% cl upper bound for @xmath139 obtained in the exo experiment taking into account nuclear matrix element uncertainties @xcite ( green band ) . the dark - green band corresponds to the @xmath111 klapdor - kleingrothaus et al . range of @xmath139 @xcite . ] let us consider the `` no - cancellation '' case , in which @xmath146 in this case , as shown in fig . [ fig : mbb - chi ] , large values of @xmath138 are excluded by the currently most stringent upper bound for @xmath139 obtained in the exo experiment @xcite ( the vertical green band in fig . [ fig : mbb - chi ] is the 90% cl exo bound taking into account nuclear matrix element uncertainties ) . this limit implies the upper bound on @xmath147 as a function of @xmath28 shown in fig . [ fig : fit - mbb ] . one can see that parts of the high-@xmath117 regions allowed by the global fit are disfavored by the exo bound . however , the large nuclear matrix element uncertainties do not allow to establish a precise bound . from fig . [ fig : fit - mbb ] one can also see that the putative klapdor - kleingrothaus et al . @xmath111 range of @xmath139 @xcite implies a rather large value of @xmath117 , around @xmath148 . in this case the oscillation length is very short , of the order of 1 cm for neutrinos with energy of the order of 1 mev , as reactor neutrinos and neutrinos emitted by radioactive sources . hence it will be practically impossible to observe a variation of the event rate characteristic of oscillations in future very short - baseline reactor neutrino experiments @xcite and radioactive source experiments @xcite ( see also @xcite ) . finally , considering cosmological measurements one must say that they are a powerful probe of the number of neutrinos and of neutrino masses at the ev scale ( see @xcite ) , but the analysis requires many assumptions on the cosmological model and its details . a comparison of the results of the fit of short - baseline oscillation data is beyond the scope of this paper . we can only say that the analysis of cosmological data in the framework of the standard @xmath149cdm @xcite allow the existence of a sterile neutrino thermalized in the early universe , but restricts its mass to be less than about 1 ev ( a possible suppression of the sterile neutrino thermalization with a large lepton asymmetry has been discussed recently in @xcite ) . if this constraint is correct , the upper bound on @xmath19 is about @xmath150 , which is much more restrictive than those of @xmath4 decay in fig . [ fig : con - mnz ] and neutrinoless double-@xmath4 decay in fig . [ fig : fit - mbb ] . in this paper we presented a complete update of the analysis of @xmath0 and @xmath1 disappearance experiments in terms of neutrino oscillations in the framework of 3 + 1 neutrino mixing . we have shown that the gallium anomaly , the reactor anomaly , solar neutrino data and @xmath2 scattering data are compatible with short - baseline oscillations with an amplitude @xmath28 between about 0.03 and 0.2 and a squared - mass difference @xmath117 larger than about @xmath151 at 95% cl . assuming the mass hierarchy in eq . ( [ mass - hierarchy ] ) , we have shown that the heavy neutrino mass @xmath129 is observable in @xmath4-decay experiments and neutrinoless double-@xmath4 decay experiments . the very recent mainz @xmath4-decay data @xcite constrain @xmath117 to be smaller than about @xmath137 at 95% cl . for majorana neutrinos , the recent exo limit on the effective majorana mass in neutrinoless double-@xmath4 decay @xcite give a more stringent constraint which can vary between about @xmath152 and @xmath153 depending on the nuclear matrix element uncertainties if there are no cancellations between the contribution of @xmath49 and that of the three light neutrinos . we think that our results are interesting for the many projects which will search in the next years effects of light sterile neutrinos with electron neutrino and antineutrino radioactive sources ( see @xcite ) , reactor electron antineutrinos ( see @xcite ) and accelerator electron neutrinos @xcite . the work of y.f . li is supported in part by the national natural science foundation of china under grant no . 11135009 . the work of h.w . long is supported in part by the national natural science foundation of china under grant no .	we present a complete update of the analysis of @xmath0 and @xmath1 disappearance experiments in terms of neutrino oscillations in the framework of 3 + 1 neutrino mixing , taking into account the gallium anomaly , the reactor anomaly , solar neutrino data and @xmath2 scattering data . we discuss the implications of a recent @xmath3 measurement which give information on the neutrino cross section in gallium experiments . we discuss the solar bound on active - sterile mixing and present our numerical results . we discuss the connection between the results of the fit of neutrino oscillation data and the heavy neutrino mass effects in @xmath4-decay experiments ( considering new mainz data ) and neutrinoless double-@xmath4 decay experiments ( considering the recent exo results ) .
the conventional understanding about heavy - ion collisions is that when there is a hard scattering of partons , a high-@xmath0 jet is created and one studies the correlation of hadrons either within the same jet or on opposite sides . the unconventional scenario is that it is possible to detect a high-@xmath0 hadron without an identifiable high-@xmath0 jet , and thus there would be no correlation of particles above the background . both possibilities will be covered in this review . the former is important because that is how we learn about the properties of the hot and dense medium through jet quenching . the latter cautions us from regarding every high-@xmath0 particle as being a part of a jet . whether correlation exists or not depends on the hadron species , the medium from which they are produced , the kinematical region where they are detected , etc . but most important of all is the dynamical properties of the partons just before and during hadronization . thus both possibilities are essential to a complete understanding of the medium created in heavy - ion collisions . for jet correlation we consider near- and far - side correlations in @xmath0 , @xmath3 and @xmath2 , as well as auto - correlation . for no jet correlation there are three areas where the phenomenon is expected to occur : @xmath2 and @xmath4 production , @xmath5 and @xmath6 production at large @xmath3 , and at large @xmath0 at lhc . only ideas and results will be presented here , quantitative details being referred to the original papers . since most of the existing data on jet correlation are for @xmath7 gev / c , and for @xmath8 even lower , fragmentation is not a reliable mechanism for hadronization at those intermediate momentum ranges @xcite . most of the quantitative calculations that can reproduce the data are based on the recombination model ( rm ) . we give here a brief overview of such studies . the @xmath0 distribution of associated particles in the same jet that contains the trigger particle has been calculated in the rm , where the thermal - shower parton recombination plays the dominant role @xcite . the results are shown in fig . 1 for both ( a ) @xmath9 and ( b ) @xmath10 collisions at 200 gev at various centralities . evidently , there is little dependence on centrality in @xmath9 collisions , but significant dependence in @xmath10 collisions . those results are in accord with the data . 2 shows the recent data from star presented at hp06 @xcite . the solid line for @xmath11 correlation @xcite agrees well with the square points for @xmath12 correlation in @xmath10 collisions . on the away side there are data for @xmath13 gev / c and @xmath14 gev / c from star @xcite as shown in fig . 3 . from the ratio of @xmath15 shown in ( d ) it is clear that there is suppression at @xmath16 gev / c but enhancement at @xmath17 gev / c . that is quite understandable as the consequence of jet quenching in the higher @xmath8 region and soft - parton regeneration in the lower region . so far there has been no quantitative theoretical calculation that reproduces these properties of the data . gev / c and @xmath18 gev / c @xcite.,width=188 ] for jet tomography it is necessary to go to higher trigger momentum , where the medium modified fragmentation function can be more reliably applied @xcite . star has data with @xmath19 in the 8 - 15 gev / c range @xcite . the away - side jets show diminishing peak heights as @xmath8 is increased . to quantify that behavior the trigger - normalized fragmentation function @xmath20 is studied , where @xmath21 is the trigger - normalized momentum fraction @xmath22 , proposed by wang @xcite . shown in fig . 4 is @xmath20 vs @xmath21 on the away side for @xmath23 @xcite . the straight lines are drawn to indicate the universality of the @xmath21 dependence among @xmath24 and @xmath25 collisions , claimed by star , if the points at @xmath26 are ignored . it seems that if the lowest point for @xmath9 min bias is also ignored , the best fit for that collision system would be a straight line with a lower slope , in comparison with which the data points for the @xmath25 system would fall off faster with @xmath21 . to focus on the ratio @xmath27 , as shown in the lower panel of fig . 4 , may be misleading , since all three @xmath20 deviate strongly from linear extrapolations at @xmath28 , while their ratios reveal no such deviation . our point here is to register an alert for the possible transition of the hadronization mechanism between @xmath29 and @xmath30 regions . the higher @xmath21 region is dominated by fragmentation and exhibits universality , while the lower @xmath21 region shows signs of enhancement due to the contribution from recombination . we suggest that the @xmath31 ratio should be measured on the away side . its dependence on @xmath21 would reveal the proportion of the two mechanisms : the larger @xmath32 is , the more is the participation of recombination , which is a property of the rm that has been established in the study of single - particle distributions @xcite . turning next to the correlation in @xmath3 and @xmath2 on the near side , we recall the characteristic of the star data @xcite that show a peak in @xmath35 sitting on top of a pedestal that stretches over the interval @xmath36 . the pedestal may well be influenced by the effect of longitudinal expansion @xcite . in the rm we have been able to reproduce both the peak and the pedestal , as shown in fig . 5 @xcite , by relating the thermalization of the soft partons generated by the energy loss of the hard parton to the pedestal with @xmath37 mev . the peak is due to the recombination of the secondary shower parton with the thermal parton , the primary shower parton being responsible for the trigger particle . the prediction for the distribution in @xmath38 agrees well with the data @xcite . at higher trigger momentum the hard partons originate closer to the surface , so less energy is lost to the medium , and the pedestal phenomenon goes away . compared to the near - side correlation , there are far more theoretical studies done on the @xmath38 distribution on the away side , since that is where the properties of hard parton propagating through a dense medium are revealed . various suggestions have been made , such as mach cone , cherenkov gluons and radiation , color wake , jet quenching and fragmentation @xcite-@xcite . most of the suggestions describe the collective response of the medium to the passage of a hard parton . such a response , especially in the case of mach cone , would result in enhancement of hadron production on both side of the recoil jet in every triggered event so that the @xmath38 distribution would exhibit a double bump structure around @xmath39 , as has been observed for @xmath40 and @xmath41 gev / c @xcite . an alternative approach to the problem is presented by chiu at this meeting @xcite . a parton multiple scattering model is constructed to reproduce the dip - bump structure in the data of @xcite by allowing the recoil parton to undergo multiple scattering as it traverses the dense and expanding medium with random scattering angle in the forward cone at each step . the parton loses energy at each step so it may be absorbed by the medium if the trajectory has a long path length . the exit tracks have the characteristics that they bend persistently in one direction and leave the medium in a few steps . while each track in this semi - classical description may not be realistic , the general features after averaging over all tracks present a sensible modeling of deflected jets . the main phenomenological feature is that there is at most one deflected jet per triggered event , but the event - averaged @xmath38 distribution exhibits the symmetrical dip - bump structure around @xmath42 , as shown in fig . the dashed line is the pedestal due to thermalized soft partons generated by energy loss . the solid line above the pedestal describes the hadron distribution due to the exit partons . when the event generator is applied to the case with higher trigger momentum in the range @xmath43 gev / c @xcite , there is only one peak on the away side at @xmath39 , with its height and width in agreement with the data shown in fig . 7 for ( a ) @xmath44 gev / c and ( b ) @xmath45 gev / c . distribution on the away side . data are from @xcite . dashed line is the calculated pedestal , and solid line includes the deflected jets @xcite.,width=302 ] distribution on the away side for @xmath46 gev / c and ( a ) @xmath47 gev / c and ( b ) @xmath48 gev / c.,width=302 ] the advantage of studying autocorrelation is that two particles are treated on the same footing and that a larger set of angular relationships between two shower partons in the same jet is included in the study of correlation compared to the analysis in which one particle is treated as a trigger and the other as the associated particle . the usual correlation function ( cumulant ) , @xmath49 , is defined in terms of the momenta @xmath50 and @xmath51 of the two particles , which in turn can be expressed in terms of angular differences and sums @xmath52 , ignoring their magnitudes if the emphasis is on @xmath33 and @xmath34 analysis . autocorrelation @xmath53 is the integral of @xmath54 over @xmath55 and @xmath56 . no ambiguous subtraction procedure is used . data from star have been analyzed yielding a wealth of information on autocorrelation @xcite . and @xmath57 @xcite.,width=226 ] in the rm autocorrelation is studied by considering two shower partons in a jet oriented in all possible directions in the jet cone . they hadronize by recombination with thermal partons for @xmath0 in the intermediate region @xcite . in fig . 8 is shown the normalized autocorrelation function for the width of the jet cone being @xmath58 . of course , other values of @xmath59 can be considered , resulting in different shapes of @xmath53 . that is where one envisions the possibility of extracting the width of the jet cone from the data by comparing the structure in fig . 8 , or its projections to either @xmath60 or @xmath61 , with similar experimental plots . so far the data on autocorrelation is for @xmath0 too low to be used meaningfully in the above sense , since there is no significant contribution of shower partons at low @xmath0 . before we go into specific problems that depart from the conventional situation where correlation always accompanies hadrons produced at high @xmath0 , let us discuss in general the physics of hadron correlation at high @xmath0 . as we have seen in the previous section , when an event is triggered by a high @xmath0 particle , it is usually due to a jet that supplies not only the trigger particle , but also other particles in the same jet , as well as a recoil jet that sends particles in roughly the opposite direction . thus the detection of high-@xmath0 particles is a necessary consequence of high-@xmath0 partons , but is it sufficient ? as for how high is `` high @xmath0 '' , we first consider the case where @xmath0 is around 4 gev / c at rhic , and later between 10 and 20 gev / c at lhc . it has been shown in @xcite that for single - particle inclusive distribution in the @xmath62 gev / c range the dominant mechanism of pion production in @xmath63 collision at 200 gev is the recombination of shower partons with thermal partons . that is the way that the medium effect is taken into account in the rm for the production of high-@xmath0 pions and protons . the recombination of thermal partons among themselves is important at lower @xmath0 . the shower partons are generated by high @xmath0 partons in hard scattering , so indeed the pions and protons produced at high @xmath0 are the necessary consequences of high-@xmath0 partons . but what if the hard scattering processes are suppressed for dynamical and kinematical reasons ? then in the absence of shower partons the thermal partons would emerge as the only partonic source for hadronization even for particles at @xmath64 gev / c or higher . although the consideration here is only for single - particle distribution , it is clear that the nature of correlation would be drastically altered when the suppression of hard scattering removes the dynamical origin of high-@xmath0 correlation of the usual kind . indeed , one should not expect any correlation if the particles produced are of thermal origin . the scenario described above turns out to show up in two cases : @xmath2 and @xmath4 production at mid - rapidity , and @xmath6 and @xmath5 production at forward rapidity . another special case involves not so much the lack of hard scattering , but too many of them at lhc . in that case correlation can not stand out above the background , so the detection of dynamical correlation should yield null result . since @xmath2 and @xmath4 are made of strange quarks only , if they are to be produced by hard scattering , the relevant processes involve either ( a ) a high-@xmath0 non - strange parton that creates an @xmath65 shower parton , or ( b ) a high-@xmath0 @xmath65 parton by direct scattering . in the former case the @xmath66 showering process is suppressed , while in the latter case the direct production of high-@xmath0 @xmath65 quark is suppressed . in either case the distribution of @xmath65 shower parton is weak , so thermal - shower recombination is not important in the intermediate @xmath0 region . that leaves only the thermal @xmath65 partons to recombine to form @xmath2 and @xmath4 . in fig . 9 are shown the @xmath0 distributions of @xmath2 and @xmath4 for all combinations of thermal @xmath67 and shower @xmath68 recombination @xcite , assuming @xmath69-function for the recombination functions . notice how much lower the contributions involving @xmath70 are compared to the dashed lines that are nearly completely covered by the solid lines that represent the sums . thus in the intermediate @xmath0 region those distributions are approximately exponential , which reveals he thermal nature of the recombining partons . the agreement with the existing data in feb 2006 is excellent . subsequently , star has extended those data to higher @xmath0 , as shown in fig . 10 @xcite . it is evident that the @xmath2 points at high-@xmath0 deviates from the straight - line extrapolation of @xcite , although the @xmath4 points seem to retain the exponential behavior out to 5.5 gev / c . the theoretical reason for the up - bending of @xmath2 is currently under investigation . the ratio @xmath71 being linearly rising up to @xmath72 gev / c is nevertheless very remarkable . the linearity of the line in the inset is due to dimensional differences in the equations for 2 and 3-quark recombination of thermal partons , but the significance is in the thermal nature of the partons . thus we find that the @xmath65 quarks are thermal out to 2 gev / c , a range of validity far wider than one normally associates with statistical distribution for @xmath73 gev / c . one may infer from this result that the medium has the characteristics of a quark - gluon plasma , which are : ( a ) strange quarks are enhanced , and ( b ) they are thermalized . now let us return to the issue about correlation . since figs . 9 and 10 indicate that the spectrum of @xmath4 is exponential out to @xmath74 gev / c , a property that is interpreted in the rm as being of thermal origin without the involvement of hard partons , then there can be no hadrons correlated with the observed @xmath4 up to 6 gev / c that are not in the background @xcite . this prediction is currently under active investigation by star . the production of hadrons at large @xmath3 is another situation where hard scattering is suppressed . if the hadronization process is fragmentation , then a hard parton must first be scattered to nearly the kinematical boundary in order for it to fragment to a hadron at large @xmath3 . but since such hard scattering must vanish at the kinematical boundary , it is highly suppressed just inside the boundary . on the other hand , hadronization by recombination involves the additivity of parton momenta so nothing needs to vanish ; indeed , the distribution of produced hadron can vary smoothly across the kinematical boundary separating the fragmentation region ( fr , @xmath75 ) and the trans - fragmentation region ( tfr , @xmath76 ) @xcite . for example , three quarks at @xmath77 can recombine and form a baryon at @xmath78 . that is not possible in @xmath79 collisions , since all three forward moving quarks must come from the projectile proton . but in @xmath80 collisions the three quarks can originate from different nucleons in the projectile nucleus , so each having @xmath77 is not forbidden . there are two complications in forward production at moderate @xmath0 in @xmath80 collisions that are absent at high @xmath0 and mid - rapidity . one is the degradation of longitudinal momentum that has been referred to as `` baryon stopping '' in an archaic language . the other is the regeneration of soft partons due to energy loss by the leading partons . the former has been studied in detail in @xmath79 collisions @xcite and then in @xmath80 collisions @xcite . the latter is considered in @xcite . the regeneration of soft partons does not matter too much to the production rate of forward proton , for which valence quarks dominate , but its effect on pion production is significant . since these are non - perturbative processes , the calculations are model dependent and data guided . the most recent data that are relevant are from brahms for @xmath81 at @xmath82 gev @xcite , where the @xmath0 distribution is given . with both @xmath3 and @xmath0 known , it is possible to determine @xmath83 . unfortunately , just as @xmath83 approaches 1 , the analysis of the @xmath0 distribution is stopped , although events have been detected with some particles having @xmath84 . nevertheless , for @xmath0 up to 2.2 gev / c , the charged particle spectrum shows exponential behavior @xcite ; it corresponds to @xmath83 in the mid - fragmentation region between 0.4 and 0.9 . our study in the rm has been able to reproduce both the @xmath0 distribution and the @xmath31 ratio @xcite . the results are shown in figs . 11 and 12 . there are no shower partons contributing because hard scattering is suppressed in the fr . the solid line in fig . 10 for @xmath85/2 $ ] that includes @xmath86 and @xmath87 approximates the average charged particles , based on @xmath5 and @xmath6 distributions that are determined by the recombination of thermal partons only , including the regenerated soft partons . the agreement with data is good . note that the @xmath31 ratio in fig . 12 reaches as high as 2 at @xmath88 gev / c . that is impossible by parton fragmentation , but in recombination it is actually difficult to keep it as low as 2 unless there is substantial regeneration of antiquarks . the reason is that valence quarks dominate in the fr so the formation of proton by recombination is much easier than to form pion due to the scarcity of antiquarks whose density depends sensitively on the regeneration effect . the degradation factor @xmath89 has been adjusted to reproduce the data @xcite . since no shower partons are involved in the above calculations , no hard partons are needed for @xmath92 gev / c , neither are they available due to the suppression of hard scattering in the forward region . thus there are no jets , and one should not expect to observe any hadrons correlated at any @xmath0 . at lhc it is predicted that there can be a large number of jets , yet no jet structure is likely to be observed @xcite . that is very peculiar , and therefore interesting to verify . the reasoning is very nearly the opposite of those in the previous two subsections . it is anticipated that the jet multiplicity is high at lhc , roughly 100 in @xmath93 . since the shower partons generated by a high-@xmath0 parton are distributed in a jet cone of non - negligible size , the overlap of jet cones of neighboring jets is unavoidable when the density of jets is high . the probability of that overlap may be difficult to calculate , but given two or three colinear partons arising from two adjacent jets , their recombination to form pion and proton is easy to calculate in the rm . thus if we allow the overlap probability @xmath94 to vary over a wide range , from @xmath95 to @xmath96 , we can determine the range of single - particle distribution of @xmath6 and @xmath5 in the region @xmath97 gev / c , high enough to leave out the thermal partons , and low enough to have high jet density . in @xcite such a calculation has been done with the results shown in fig . the parameter @xmath98 refers to the suppression factor of hard partons emerging from the super - dense medium , and is related to @xmath99 . at rhic it is @xmath100 @xcite . since it is unknown at lhc , two reasonable values are used in fig . 13 . the more important dependence of the distributions is on @xmath94 . the pion distributions in the left panel reach their limiting behavior of no 2-jet recombination at @xmath101 , but the proton distributions on the right continue to decrease in magnitude at small @xmath94 . the interpretation is that proton is so much more abundantly produced because of 3 quark recombination that even at @xmath101 the 2-jet contribution is larger than the 1-jet contribution , which is for @xmath102 ( not shown ) . since it is expected that @xmath94 decreases with @xmath0 , a form @xmath103 is used to see how the distributions approach the 1-jet limit at high @xmath0 . that is shown by the heavy lines in fig . for @xmath6 they approach the limiting curves before @xmath0 reaches 20 gev / c , but for @xmath5 they do not . since for @xmath104 gev / c the average jet multiplicity is less than 2 in @xmath105 , @xmath106 is expected to decrease much faster than @xmath107 , so the heavy line for @xmath5 should rapidly join the @xmath102 line beyond . the important phenomenological consequence of the above results is the ratio @xmath32 that is insensitive to @xmath98 . that ratio is shown in fig . 14 . for fixed @xmath94 the curves increase with @xmath0 , reflecting the dominance of 3-quark recombination . for @xmath103 , the heavy lines show that @xmath32 decreases from 20 to 5 , but is significantly higher than that of the one - jet case , shown by the light solid lines in the same figure . such a large @xmath31 ratio at such a high region of @xmath0 is a result that differs spectacularly from what one expects from fragmentation in the conventional approach to high-@xmath0 physics . in addition to the surprisingly large @xmath31 ratio , there is also the unusual correlation characteristics . at rhic , where there is a jet , there are correlated hadrons . but at lhc so many jets are produced , they are part of the background . since hadrons in the @xmath109 gev / c range are not the fragments of any super - hard parton at much higher @xmath0 , but are the recombination products of semihard partons at lower @xmath0 that are numerous in every event , treating a detected hadron as a trigger does not eliminate any other hadrons formed by other jets in the same event . that is , there are many associated particles , all of which are part of the background . thus it is not possible to find any correlated particles distinguishable from the background @xcite . that property is independent of @xmath34 , since recoil partons are weaker than the ambient jets produced near the surface on the far side . data on these predictions can provide crucial guidance to the understanding of high - density medium whose emitted particles at high @xmath0 has usually been described in a framework that may have to be modified .	an overview of the correlation among high-@xmath0 hadrons produced in heavy - ion collisions is presented . emphasis is placed on the physical processes that can quantitatively account for the data on correlations in @xmath1 and @xmath2 on the near and far sides . predictions are made on processes that have no observable correlations for hadrons produced at intermediate and higher @xmath0 at rhic and lhc .
the study of code loops began with the paper @xcite of griess . let s recall the definition of code loop . first let @xmath2 be a @xmath3-dimensional vector space over a field with two elements @xmath4 . for vectors @xmath5 and @xmath6 we define @xmath7 and @xmath8 a double even code is a subspace @xmath9 such that @xmath10 and @xmath11 for all @xmath12 . let @xmath13 be a double even code and @xmath14 be the set @xmath15 . in @xcite r.griess proved that there exists a function @xmath16 , called factor set , such that @xmath17 where @xmath18 denotes the number of positions in which the coordinates of @xmath19 , @xmath20 and @xmath21 are both nonzero . in order to define code loop , let @xmath22 and @xmath23 , and let @xmath24 denotes @xmath20 and @xmath25 denotes @xmath26 . we define a product @xmath27 em @xmath14 by : @xmath28 with the product defined above , griess ( 1986 , @xcite ) proved that @xmath29 , or merely @xmath14 , has a moufang loop structure . the moufang loop @xmath14 is called code loop . we say that @xmath14 has rank @xmath30 , if dim@xmath31 . moreover , chein and goodaire ( 1990 , @xcite ) proved that code loops have a unique nonidentity square , a unique nonidentity commutador , and a unique nonidentity associator . in other words , for any @xmath32 : @xmath33 & = u^{-1}v^{-1}uv = ( -1)^{\frac{\|u \cap v\|}{2}}0 , \nonumber\\ ( u , v , w)&=((uv)w ) ( ( u(vw))^{-1})=(-1)^{\|u\cap v\cap w\|}0 . \end{aligned}\ ] ] on the other hand , a moufang loop @xmath34 is called * @xmath35-loop * if there is a central subloop @xmath36 with @xmath37 elements such that @xmath38 , where @xmath39 is the variety of groups with identity @xmath40 chein e goodaire ( 1990 , @xcite ) proved that finite code loops may be characterized as moufang loops @xmath34 for which @xmath41 . so , directly from the proof of this result we can obtain that a finite moufang loop @xmath34 is a code loop if and only if @xmath34 is a @xmath35-loop . in this paper , we study the representations of code loops . in section 2 , we prove that code loops of rank @xmath3 can be characterized as a homomorphic image of a certain free moufang loops with @xmath3 generators and we introduce the concept of characteristic vectors associated with code loops . in section 3 , we present the classification of code loops of rank @xmath0 and @xmath1 and their corresponding groups of outer automorphisms . there are exactly 5 nonassociative code loops of rank @xmath0 ( up to isomorphism ) and 16 nonassociative code loops of rank @xmath1 ( up to isomorphism ) . by using this , in section 4 , we determine all basic representations of code loops of rank @xmath0 and @xmath1 . in the following we introduce the variety of moufang loops @xmath43 , generated by all code loops . let @xmath43 be the variety of moufang loops with the following identities : @xmath44 ^ 2 = 1 , \ , ( x , y , z)^2 = 1 , \nonumber \\ & \left[x^2,y \right ] = 1,\ , \left[\left[x , y\right],t \right ] = 1,\ , \left[(x , y , z),t \right ] = 1 , \nonumber \\ & ( x^2,y , z ) = 1 , \ , ( \left[x , y \right],z , t ) = 1 , \;\ ; ( ( x , y , z),t , s)=1 . \label{eq1}\end{aligned}\ ] ] we observe that code loops are contained in @xmath43 . in fact , a code loop is a moufang loop with a unique nonidentity commutator ( associator , square ) . besides , chein and goodaire ( @xcite , theorem 1 ) proved that if @xmath34 is a moufang loop with a unique nonidentity square @xmath45 , then @xmath46 and either @xmath34 is an abelian group or else @xmath47=(l , l , l)=l^{2}=\{1,e\}\subseteq { \cal z}(l)$ ] . here , @xmath47 $ ] , @xmath48 , and @xmath49 denotes , respectively , the sets of all commutators @xmath50 $ ] , all associators @xmath51 and all squares @xmath52 , @xmath53 . besides , the center of @xmath34 , @xmath54 , is the set of elements of @xmath34 which associate with every pair of elements of @xmath34 and which commute with every element of @xmath34 . note that those identities are not independents . it is interesting problem to find the minimal set of identities that defines the variety @xmath43 ( see the conjecture below ) . [ cn1 ] the variety @xmath43 has the following minimal set of identities : @xmath55 ^ 2 = 1 , \left[x^2,y \right ] = 1,(x^2,y , z ) = 1.\ ] ] let @xmath13 be a @xmath56-space with a basis @xmath57 . we identify each element of @xmath13 with a finite subset of the set of natural numbers @xmath58 as follows : we consider the biunivocal correspondence @xmath59 , where @xmath60 in this way , we use the notation @xmath61 when @xmath62 , for @xmath63 . besides , we observe that , @xmath64 , where @xmath65 . we consider @xmath66 and @xmath67 antisymmetric products of @xmath13 . this means that @xmath68 has a basis @xmath69 and @xmath70 has a basis @xmath71 . moreover , @xmath72 if and only if @xmath73 . let @xmath74 be an isomorphic copy of @xmath13 with a basis @xmath75 let the set @xmath76 . we want to define a product @xmath77 such that for @xmath78 we have @xmath79=i^{-1}j^{-1}ij = i\wedge j , \;\;(i , j , k ) = ( ij.k)(i.jk)^{-1}= i\wedge j\wedge k . \label{eq2}\end{aligned}\ ] ] for the following , given an element @xmath80 , we identify it in @xmath81 with a product @xmath82 . now we define @xmath83=(x , y , z)=1 , \;\;\;x\in { \bar{v}}.w.u , \;\;y , z \in \cal f. \label{eq4}\end{gathered}\ ] ] by definition @xmath84 . now let @xmath85 be of the form @xmath86 and @xmath87 , where @xmath88 and @xmath89 . therefore , the product of @xmath20 by @xmath21 is given by : @xmath90 . for the following proofs we use the notation more concise , for all @xmath91 : @xmath92&=\displaystyle\prod_{i \in \sigma , j \in \mu , k \in \lambda , j < k } { i\wedge j\wedge k } \label{eq5}\\ \left[\sigma,\mu\right]&=\displaystyle\prod_{i \in \sigma , j \in \mu , i > j } { i\wedge j } \label{eq6}\\ \{\sigma,\mu\}&=\left\{\begin{array}{c } \displaystyle\prod_{i \in \sigma \cap \mu } { \bar{i } } , \;\;\;\mbox{se } \;\;\;\sigma \cap \mu \neq \emptyset\\ 1 , \;\;\;\mbox{se } \;\;\;\;\;\;\sigma \cap \mu = \emptyset \label{eq7 } \end{array}\right.\end{aligned}\ ] ] [ lemma2.3 ] with the above definitions , in @xmath81 are valid the following equalities , for all @xmath91 : @xmath93[\lambda,\sigma][\sigma \delta \mu,\lambda \delta \sigma]=[\mu,\lambda][\sigma,\lambda \delta \mu][\sigma \delta \mu \delta \lambda,\sigma]\label{eq9}\\ \left[\lambda \delta \tau , \mu , \sigma\right]=[\lambda , \mu , \sigma][\tau , \mu , \sigma ] \nonumber\\ \left[\sigma , \lambda \delta \tau , \mu\right]=[\sigma , \lambda , \mu][\sigma , \tau , \mu ] \nonumber\\ \left[\sigma , \mu , \lambda \delta \tau\right]=[\sigma , \mu , \lambda][\sigma , \mu , \tau ] \label{eq10}\\ \left[\sigma , \lambda , \sigma\right][\sigma , \sigma , \lambda]=1 \label{eq11}\\ \left[\sigma , \lambda , \mu\right][\mu , \lambda,\sigma][\sigma,\mu , \lambda][\mu , \sigma , \lambda]=1 \label{eq12}\\ \left[\sigma \delta \mu,\lambda \delta \sigma , \lambda \delta \sigma\right][\sigma,\mu,\mu][\lambda , \sigma , \sigma][\sigma \delta \mu \delta \lambda , \sigma , \sigma][\sigma , \mu \delta \lambda , \mu \delta \lambda][\mu,\lambda , \lambda]=1 \label{eq13}\end{gathered}\ ] ] we denote @xmath94 we write @xmath95 , [ ij , k ] , [ i , j , k]$ ] and @xmath96$]instead @xmath97 , [ \xi_{i } \cup \xi_{j } , \xi_{k } ] , [ \xi_{i } , \xi_{j } , \xi_{k } ] $ ] and @xmath98 $ ] , respectively . to prove the equality ( [ eq8 ] ) , first we see that @xmath99 thus , by the equality ( [ eq14 ] ) and by the fact @xmath100 for all @xmath101 , we can rewrite the right part of ( [ eq8 ] ) as follows : @xmath102 in fact : 1 . [ item1 ] @xmath103 + we note that @xmath104 , then @xmath105 . analogously , @xmath106 . we also note that @xmath107 and @xmath108 , so we have @xmath109 therefore , @xmath110 2 . [ item2 ] @xmath111 3 . [ item3 ] since @xmath112 , then @xmath113 thus , by ( [ item1]),([item2 ] ) and ( [ item3 ] ) , we have @xmath114 analogously , we prove that the left part of ( [ eq8 ] ) can be rewritten in the same way . now we are going to prove the equality ( [ eq9 ] ) . we have @xmath115=[\sigma,\lambda][\mu,\lambda]$ ] , if @xmath116 . by the diagram ( fig.[diagram ] ) , we can rewrite @xmath117 and @xmath118 as follows : @xmath119 , @xmath120 and @xmath121 . so the right and left parts of ( [ eq9 ] ) can be rewritten as a product of @xmath122 $ ] , @xmath123 . as @xmath122^{2}=1 $ ] , the equality ( [ eq9 ] ) is valid . to prove the firt relation in ( [ eq10 ] ) , we note that @xmath124=\left[\lambda,\mu , \sigma\right]\left[\tau , \mu , \sigma\right],$ ] if @xmath125 therefore , @xmath126[\lambda \cap \tau , \mu , \sigma][\lambda \cap \tau , \mu , \sigma][\tau \setminus \lambda , \mu , \sigma ] = [ \lambda , \mu , \sigma][\tau , \mu , \sigma ] . \end{aligned}\ ] ] analogously , we prove @xmath127=[\sigma , \lambda , \mu][\sigma , \tau , \mu]$ ] and @xmath128=[\sigma , \mu , \lambda][\sigma , \mu , \tau ] \nonumber$ ] . we are going to prove ( [ eq11 ] ) and ( [ eq12 ] ) in the cases @xmath129 . first we consider @xmath130 and @xmath131 . suppose @xmath132 . we have to analyze three cases : @xmath133 , @xmath134 and @xmath135 . in the first case , the product @xmath136 appears twice in @xmath137[\sigma,\sigma,\lambda]$ ] , as a factor of @xmath137 $ ] and as a factor of @xmath138 $ ] . in the second case we have a factor of the form @xmath139 in @xmath138 $ ] and there are not nontrivial factors in @xmath137 $ ] with @xmath140 , while in the third case we have the factor @xmath141 in @xmath137 $ ] and there are not nontrivial factors in @xmath138 $ ] , with @xmath140 . those cases where @xmath142 are analogous . therefore , the relation ( [ eq11 ] ) is valid in the particular case @xmath143 . now let @xmath140 be elements of @xmath144 and @xmath145 respectively . we consider the same three cases to prove the relation ( [ eq12 ] ) in the particular case . in any case we will have the factor @xmath146 exactly twice in @xmath147[\mu , \lambda,\sigma][\sigma,\mu , \lambda][\mu , \sigma , \lambda]$ ] . for example , in the first case there are not factors with @xmath140 in @xmath148 $ ] and @xmath149 $ ] , and there is one factor @xmath136 in @xmath150 $ ] and one factor @xmath151 in @xmath152 $ ] . to prove the relation ( [ eq11 ] ) in general case , we consider again @xmath153 and @xmath154 . the term @xmath137=[1324,1357,1324]$ ] can be rewritten as @xmath155[13,13,24][13,57,13][13,57,24][24,13,13][24,13,24][24,57,13][24,57,24].\ ] ] in the same way the term @xmath138=[1324,1324,1357]$ ] can be rewritten as @xmath155[13,13,57][13,24,13][13,24,57][24,13,13][24,13,57][24,24,13][24,24,57].\ ] ] by definition , @xmath156 . so we obtain the relation ( [ eq11 ] ) in the general case . next , by the relation ( [ eq11 ] ) we obtain that @xmath157[\sigma \delta \mu , \sigma \delta \mu , \lambda]=1 $ ] and by the relation ( [ eq10 ] ) we prove the relation ( [ eq12 ] ) in the general case . finally , the relation ( [ eq13 ] ) is obtained directly from relations ( [ eq10 ] ) , ( [ eq11 ] ) and ( [ eq12 ] ) . the set @xmath81 defined above is a moufang loop in the variety @xmath158 . by the definition of @xmath159 we have @xmath160 . we consider @xmath91 . we are going to prove the following moufang identity : @xmath161 by the definition ( [ eq3 ] ) we have : @xmath162.\left[\sigma,\mu , \mu \right]).((\lambda \delta \sigma).\{\lambda,\sigma\}.\left[\lambda,\sigma \right].\left[\lambda,\sigma , \sigma \right])\ ] ] and by ( [ eq3 ] ) and ( [ eq4 ] ) we have : @xmath163.\left[\sigma,\mu \right].\left[\lambda,\sigma \right].\left[\sigma,\mu , \mu \right].\\&&\left[\sigma \delta \mu,\lambda \delta \sigma,\lambda \delta \sigma\right].\left[\lambda,\sigma , \sigma \right]\end{aligned}\ ] ] on the other hand , @xmath164.\left[\mu , \lambda , \lambda\right]).\sigma\\&=&(\sigma \delta ( \mu \delta \lambda).\{\sigma , \mu \delta \lambda\}.\{\mu , \lambda\}.\left[\mu , \lambda\right].\\&&\left[\sigma , \mu \delta \lambda\right].\left[\sigma , \mu \delta \lambda , \mu \delta \lambda\right].\left[\mu , \lambda , \lambda\right]).\sigma\\&=&(\mu \delta \lambda).\{\sigma \delta \mu \delta \lambda , \sigma\}.\{\sigma , \mu \delta \lambda\}.\{\mu , \lambda\}.\\ & & \left[\sigma \delta \mu \delta \lambda , \sigma\right].\left[\sigma , \mu \delta \lambda\right].\left[\mu , \lambda\right].\left[\mu,\lambda,\lambda\right].\\&&\left[\sigma \delta \mu \delta \lambda , \sigma , \sigma\right].\left[\sigma , \mu \delta \lambda , \mu \delta \lambda\right]\end{aligned}\ ] ] now the equality ( [ eq15 ] ) follows from the lemma [ lemma2.3 ] . we consider a subloop of @xmath81 , denoted by @xmath165 and with generators @xmath166 , where @xmath167 . the next proposition contains important results that we are going to use throughout this paper . the demonstration of the first and second result comes directly from ( @xcite , theorem 2 ) . the last result is proved easily . [ prop2.5]let @xmath168 be a moufang loop . 1 . if @xmath169 and all the commutators and associators of @xmath168 are central , then @xmath170=[x , z][y , z](x , y , z ) . \end{aligned}\ ] ] 2 . if the commutators and associators of @xmath168 are central , then @xmath171 3 . if the squares and commutators of a moufang loop @xmath168 are central , then @xmath172.\end{aligned}\ ] ] [ lemma2.6 ] let @xmath173 be a free loop in @xmath43 , with free generator set @xmath174 . then for all @xmath175 , there are @xmath176 , @xmath177 , @xmath178 , with @xmath179 such that @xmath180^{\xi_{ij}}.\prod_{i < j < k}(x_{i},x_{j},x_{k})^{\xi_{ijk}}. \label{eq16}\ ] ] let @xmath181 be a central subloop of @xmath182 generated by the set @xmath183,(x_{i},x_{j},x_{k})\|i , j , k\in i_n\}.$ ] by the proposition [ prop2.5 ] we have : @xmath184;\\ \left[x_{i}x_{j},x_{k}\right]&=&\left[x_{i},x_{k}\right]\left[x_{j},x_{k}\right](x_{i},x_{j},x_{k});\\ ( x_{i}x_{j},x_{k},x_{p})&=&(x_{i},x_{k},x_{p})(x_{j},x_{k},x_{p}).\end{aligned}\ ] ] hence @xmath185 is an abelian group of expoent two . then @xmath186 and the lemma is proved . [ theorem2.7]@xmath165 is a free loop in the variety @xmath43 of rank @xmath3 . let @xmath187 , @xmath188 be a free loop with @xmath3 generators , then by the definition of free loop there is a unique surjective morphism @xmath189 such that @xmath190 to show that @xmath165 is free , we just need to prove that @xmath191 is injective . it is clear that @xmath192 by the lemma [ lemma2.6 ] any @xmath193 has the form given in ( [ eq16 ] ) . then @xmath194 by definition of @xmath195 therefore , @xmath165 is free loop . \(1 ) for any code loop @xmath34 of rank @xmath3 there is a homomorphism @xmath196 such that @xmath197 and codim@xmath198 \(2 ) for all @xmath199-subspace @xmath200 of codimension @xmath201 we have a code loop @xmath202 \(3 ) the loop @xmath203 is a group if and only if @xmath204 , where @xmath205 [ prop2.9 ] let @xmath206 and @xmath207 be @xmath199-subspaces of @xmath208 of codimension @xmath201 , so @xmath209 if and only if there is an automorphism @xmath191 of @xmath165 such that @xmath210 let @xmath211 be an automorphism such that @xmath212 where @xmath206 and @xmath207 are @xmath199-subspaces of @xmath213 of codimension @xmath201 . we define @xmath214 by @xmath215 , for all @xmath216 . clearly , @xmath191 is an isomorphism between code loops . now we construct the following commutative diagram : we suppose that @xmath217 , @xmath218 is an isomorphism , @xmath219 is a surjective homomorphism and @xmath220 is the inclusion application , for @xmath221 . thus , @xmath222 & t_{j}\ar[r]^{i_{j } } & { \cal f}_{n}\ar[r]^{\pi_{j } } & l_{j } \ar[r ] & 1}\ ] ] is a short exact sequence , for @xmath221 . let @xmath223 be a set of free generators of @xmath165 and @xmath224 since @xmath225 is surjective , there exists @xmath226 such that @xmath227 then there exists a homomorphism @xmath228 such that @xmath229 so , since both the right part of the diagram is commutative by construction , we have @xmath230 we have that both @xmath206 and @xmath207 has the same dimension , then we just need to prove that @xmath191 is surjective . in fact , let @xmath231 , then @xmath232 , but @xmath233 is surjective , thus , there exists @xmath234 such that @xmath235 . on the other hand , @xmath236 , because the sequences of lines of the previous diagram are short exact sequences . hence @xmath237 , that is , @xmath238 . as the right part of the diagram is commutative , we have @xmath239 . then @xmath240 , because @xmath241 is bijective , that is , @xmath242 . hence , @xmath243 , as we wanted . we denote by @xmath244 the set of all subspaces @xmath245 of codimension @xmath201 such that @xmath246 . the group @xmath247 , group of automorphisms of @xmath248 , acts on @xmath244 . let @xmath249 be the set of the correspondents @xmath250-orbits . we denote by @xmath251 the orbit of @xmath252 , that is , @xmath253 . the correspondence @xmath254 gives a bijection between @xmath249 and the set of the isomorphism classes of nonassociative code loops of rank @xmath3 . let @xmath34 be a code loop with generator set @xmath255 and center @xmath256 . then we define the characteristic vector of @xmath34 , denoted by @xmath257 or @xmath258 , by @xmath259 where @xmath260 , @xmath261\ ; \;\mbox{e}\ ; \ ; ( -1)^{\lambda_{ijk}}=(x_{i},x_{j},x_{k})$ ] . with @xmath262 we can associate a subspace @xmath263 in @xmath208 of codimension 1 as follows : we choose an element @xmath264 and we denote it by @xmath265 . then @xmath266,x[x_{i},x_{j } ] \| \lambda_{pq}=0 , \lambda_{ij}=1\right\}$ ] + + @xmath267 , by definition . the following proposition is a simple corollary of definitions . the application @xmath268 is a bijection between the set of characteristic vectors of nonassociatives code loops of rank @xmath3 and the subspaces @xmath252 in @xmath269 of codimension 1 such that @xmath270 . so we have a @xmath250-action over the set of the characteristic vectors . the action is defined by @xmath271 if and only if @xmath272 , for every @xmath273 . without loss of generality we can assume @xmath274 . in this case , we have @xmath275 characteristic vectors where @xmath276 . let @xmath277 be a set of @xmath278orbits of characteristic vectors of the code loops of rank @xmath3 . then @xmath279 for a given code loop @xmath34 , we denote by @xmath280 the group of automorphisms of @xmath34 and we define @xmath281 , the group of outer automorphisms of @xmath34 . by definition , @xmath282 let @xmath283 we note that @xmath284 induces an automorphism of the correspondent code loop @xmath285 . for the next proposition , we define @xmath286 and we denote by @xmath287 the general linear group of degree @xmath3 over a finite field with 2 elements . let @xmath288 then : 1 . @xmath289 ; 2 . @xmath290 and 3 . @xmath291 . let @xmath292 , so @xmath293 , for @xmath294 and @xmath295 . thus , for all @xmath292 , there are @xmath296 a corresponding basis of @xmath13 and @xmath297 . we have that @xmath298 is a vector @xmath299-space and , for @xmath292 , we have an automorphism @xmath300 such that @xmath301 . then @xmath302 . we are going to show now that @xmath303 . in fact , let @xmath284 , then @xmath304 , hence @xmath305 is an isomorphism of code loops . the homomorphism @xmath306 such that @xmath307 , is surjective ( by the proposition [ prop2.9 ] ) with kernel @xmath308 . consider also the surjective homomorphism @xmath309 . then @xmath310 is a surjective homomorphism . we have @xmath311 . now , let @xmath312 . if @xmath313 , there is @xmath314 , for all @xmath315 , then for @xmath316 , we have @xmath317 @xmath318 , hence @xmath319 in @xmath320 . thus , @xmath321 and , therefore , @xmath303 . we denote @xmath322 by @xmath323 . a @xmath324space is a @xmath325space @xmath13 with an antisymmetric trilinear form @xmath326 with values in @xmath299 . a @xmath327space is a @xmath324space @xmath13 with a bilinear map @xmath328:v \times v \longrightarrow { \bf f}_{2}$ ] such that @xmath329=[wu , v]=[v , w][v , u](v , w , u).$ ] for a given code loop @xmath34 , we can associate a @xmath37-space @xmath330 with a @xmath0-form induced by the associator and the correspondent antisymmetric aplication induced by the commutator . let @xmath13 be a @xmath324space . by definition , the nucleus of @xmath13 is the subspace @xmath331 similarly , for a @xmath327space @xmath13 , we define @xmath332=1\right\}.$ ] let @xmath333 and @xmath334 be code loops . we define the product of @xmath333 and @xmath334 as a code loop @xmath335 , where @xmath336 is a diagonal of the center of @xmath337 . let @xmath34 be an other code loop , we will write @xmath338 if @xmath339 and we will say that @xmath333 and @xmath334 are @xmath34-equivalents . for the following , @xmath340 is the ( additive ) group of integers modulo m. [ lemma2.15 ] two code loops @xmath333 and @xmath334 are isomorphic as @xmath37-spaces if and only if @xmath341 . let @xmath333 and @xmath334 be @xmath342-equivalents . therefore , there exists an isomorphism @xmath343 . we define a linear map @xmath344 by @xmath345 if @xmath346 , where @xmath347 . since by definition @xmath348=(a , l_{2},l_{2})=1 $ ] , then @xmath349 is an isomorphism of @xmath333 and @xmath334 as @xmath37-spaces . conversely , if @xmath344 is an isomorphism of code loops as @xmath37-spaces , so the application @xmath350 , where @xmath351 and @xmath352 if @xmath353 , @xmath354 if @xmath355 , @xmath356 , @xmath357 , is an isomorphism . classification of all code loops untill rank 4 using the definition of characteristic vector has been done in @xcite . in this section we will illustrate the notions introduced in this case . let @xmath34 be a nonassociative code loop of rank @xmath0 with generators @xmath358 then@xmath359 . we associate to @xmath34 the characteristic vector @xmath258 , defined by@xmath360 where @xmath361,@xmath362 , @xmath363,@xmath364=(-1)^{\lambda_4}$ ] , @xmath365=(-1)^{\lambda_5}$ ] , @xmath366=(-1)^{\lambda_6}.$ ] [ theorem3.1 ] consider @xmath367 the code loops with the following characteristic vectors : @xmath368 then any two loops from the list @xmath369 are not isomorphic and all nonassociative code loop of rank @xmath0 is isomorphic to one of this list . in the first case , we can choose generators such that @xmath364 = \left[a , c\right ] = \left[b , c\right ] = 1 $ ] . in fact , if @xmath365 = -1 $ ] , we can take @xmath371 and we will have @xmath372 = \left[a , b\right]\left[a , c\right](a , b , c ) = 1,$ ] ( see theorem 2 , @xcite ) . notice that the loops that correspond to the last two characteristic vectors are isomorphics . in fact , let @xmath375 be the code loop with generator set @xmath376 and with corresponding characteristic vector @xmath377 . then , for the generator set @xmath378 , we have @xmath379 , that is , @xmath380 . i ) : : @xmath381 : here we have @xmath382 . ii ) : : @xmath383 if we assume @xmath384 or @xmath357 , we will have @xmath385 . hence we can choose generators @xmath370 such that @xmath386 . + therefore , @xmath387 let @xmath400 be a fixed set of generators for each code loop @xmath401 , @xmath402 . we consider the characteristic vector @xmath403 as representative of each orbit @xmath404 , @xmath405 . to find all characteristic vectors belonging to the orbit of each representative , we have to find the different set of generators formed from the set of generators @xmath400 fixed . in the table [ tabqccaso3 ] , we present all the values of squares and commutators obtained from each characteristic vector , that is , the squares of each generator element and the commutator between each pair of generators elements . each column @xmath406 corresponds to a code loop @xmath401 , @xmath402 .	in this work we construct free moufang loop in the variety generated by code loops . we apply this construction for study the code loops . moreover , we define and determine all basic representations of code loops of rank @xmath0 and @xmath1 .
the use of cepheids as stellar candles requires a deep knowledge of their pulsational characteristics . the separation between fundamental ( @xmath1-mode ) and first - overtone ( @xmath2-mode ) radial pulsators @xcite is still the most relevant distinction to be taken into account to calibrate the period - luminosity relations . about the regularity of their variations , ground - based photometry of v473 lyr @xcite and polaris @xcite showed continuous changes in amplitude and phase . @xcite established that v473 lyr pulsates in the second radial overtone and shows strong amplitude and phase modulations on two long periods ( 1205@xmath33 and 5300@xmath3150 d ) . the slow changes in the light curve of polaris on a time baseline of 4.5 yr were precisely recorded by space photometry using the _ coriolis _ spacecraft . @xcite connected them to a quite fast evolutionary phase . however , @xcite proposed a different interpretation pointing out how the amplitude change is cyclic rather than monotonic and also suggesting an excess of power in the 2 - 6 d range due to the granulation . the situation was expected to change with the advent of high - precision space mission . a detailed analysis of the _ most _ observations of sz tau and rt aur revealed variations from cycle to cycle as a function of the pulsation phase @xcite . the stars were both monitored for less than seven consecutive cycles : the @xmath2-cepheid sz tau showed a pulsation less stable than the @xmath1-cepheid rt aur . the historical analysis of the times of maximum brightness proved that both stars experienced period variations on a time - scale of decades and that those of the overtone mode resulted to be more erratic than the @xmath4 ones . in the case of _ kepler _ , there was only one cepheid in the field - of - view , i.e. , v1154 cyg . @xcite provided a very detailed analysis of the data acquired in the first 600 d ( i.e. , about 120 cycles ) of the mission , detecting significant cycle - to - cycle fluctuations . a very slight correlation was also found between the fourier parameters and the o - c values , i.e. , the difference between the _ observed _ and _ calculated _ times of maximum brightness ( t@xmath5 ) . this suggested that the o - c variations ( up to 30 min ) might be due to instabilities in the light curve shape . on the basis of these observational results , @xcite suggested that convection and hot spots can explain the observed period jitter , thus supporting the @xcite interpretation of the polaris changes . the clues about the changes in the light curve of v1154 cyg were strengthened by the analysis of new _ kepler _ measurements ( up to q17 quarter ) : the analysis of the fourier parameters supplied a robust determination of a periodicity of 158.6 d @xcite . about ground - based observations , @xcite discovered significant modulations in the radial velocity curves of some cepheids . these modulations appear to be different from those reported on v1154 cyg , since they manifest as shape and amplitude variations on time - scales of years for short - period cepheids ( like v1154 cyg ) and cycle - to - cycle variations in the long - period cepheids . a common result among all these investigations is that no nonradial mode was identified in the light or radial velocity curves of galactic cepheids . hence the counterparts of the subclasses detected in the large magellanic cloud ( lmc ; for a review see * ? ? ? * ) are still missing . however , an important player has not yet contributed , i.e. , the corot data base of the light curves collected in the exoplanetary fields . such data base already proved to be a real treasure allowing the discovery of a unique triple - mode cepheid in the outskirts of the milky way @xcite . in this paper we describe the new elements brought by the corot photometry to the hot topic of the stability of the cepheid light curves . the first new cepheid discovered by corot was 2mass j06415168 - 0120059@xmath6corot 0102618121 , a 12@xmath7-mag star located at @xmath8 = @xmath9@xmath10@xmath11 , @xmath12 = 012005 ( j2000 ) . it was observed during the first long run in the anticentre direction ( lra01 ) and then in the lra06 four years later ( table [ runs ] ) . the classification of the type of variability was a little troublesome . the star was assigned to the rr lyr working group , but the preliminary analysis immediately pointed out that the star was not a rr lyr variable since the period was too long . the cepheid variability was much more probable , but some doubts still remained , based on the classification as m1v type reported in the header of the fits file of the lra01 data . indeed , the corot variability classifier " automated supervised method ( cvc ; * ? ? ? * ) suggested an active , rotational variable as first hypothesis and a classical cepheid as a second one . the spectral classification of corot 0102618121 was revised as a5v in the current version of the exo - dat @xcite catalogue and consequently reported in the header of the more recent lra06 data . while the analysis of corot 0102618121 was in progress , we proposed known cepheids as corot targets by answering the announcements of opportunity regularly issued before any run . in such a way v376 mon@xmath6corot 0300003751 and dp mon@xmath6corot 0221644967 were monitored in the lra02 and sra02 runs , respectively ( table [ runs ] ) . however , it should be expected that other cepheids would be discovered thanks to the deep survey performed by corot on the galactic plane . to search for these new cepheids , we first scrutinized all the new variables discovered with the berlin exoplanet search telescope ( best ) that have been classified as candidate cepheids @xcite and actually observed with corot . we noticed that the variability of v376 mon was recorded @xcite with the correct period , i.e. , 1.652 d @xcite , not the 0.623 d reported in the discovery paper @xcite . unfortunately none of other best candidates were confirmed as a new cepheid , since they turned out to be rotational variables . the misclassification is probably due to the too relaxed criterion used to classify a new variable as cepheid , i.e. , a periodic variability with an amplitude larger than 0.10 mag . indeed , such an amplitude threshold can be reached by active spotted stars . as a second step in the data mining of the corot data base , we selected all the variables for which the cvc method proposed a classification as cepheid , independently from rank and probability . the full light curves of about 250 candidates thus selected were inspected by eye . eclipsing variables , rr lyr and high - amplitude @xmath12 sct stars were found and not considered furthermore . we remained with a sample where the dichotomy was again between cepheid and rotational variables . the stars characterized by very large cycle - to - cycle variations were immediately considered as rotational variables . the stars showing more stable light curves were analysed in frequency and least - squares solutions of the mean light curves were obtained . we note that in this process the small amplitude was not considered as a rule of thumb since it could be reduced by an unresolved , blended star . on the other hand , a large amplitude ( say , greater than 0.20 mag ) was recognized as the signature of a pulsational mechanism . most of the stars were considered as rotational variables after that an ensemble of several parameters ( period , amplitude , stability of the light - curve shape , fourier parameters , ... ) was taken into account . however , we can not rule out that a few of them are actually cepheids , whose regular curve is perturbed by that of a blended irregular variable . we could identify some borderline cases ( e.g. corot 0102635101 , 0102640143 , 0605539616 , 0223971984 , 0102578739 , 0110827792 , and , overall , 0223971984 ) where the distinction between activity and pulsation was very difficult . new spectroscopic observations are necessary to establish the nature of these problematic stars in an unambiguous way . during the screening procedure we discovered the unique galactic triple - mode cepheid corot 0223989566 @xcite and at the end we could count seven bona - fide cepheids . in addition to corot 0102618121 , v376 mon , and dp mon we identified four new cepheids : corot 0221640090 , 0315221415 , 0211626074 , and 0659466327 ( table [ runs ] ) . the corot photometry of v376 mon and of other cepheids was analysed when the interpretation of the corot 0102618121 data was almost finished . therefore , we describe the procedure as applied to corot 0102618121 first and then how it has been extended to other cepheids later . rc lrrr lr r & & & & & & & & + 0102618121 & & lra01 & oct . 23 , 2007 & mar . 3 , 2008 & 131.46 d & chr & 32 s & 326634 + 0102618121 & & lra06 & jan . 12 , 2012 & mar . 29 , 2012 & 76.61 d & chr & 32 s & 196032 + 0300003751 & v376 mon & lra02 & nov . 16 , 2008 & mar . 8 , 2009 & 111.64 d & chr & 512 s & 16400 + 0221644967 & dp mon & sra02 & oct . 11 , 2008 & nov . 12 , 2008 & 31.75 d & chr & 32 s & 72144 + 0659466327 & & lrc09 & apr 12 , 2012 & july 5 , 2012 & 83.51 d & mon & 512 s & 11894 + 0315221415 & & sra03 & mar . 5 , 2010 & mar . 29 , 2010 & 24.73 d & mon & 512 s & 3670 + 0211626074 & & src01 & apr . 13 , 2007 & may 9 , 2007 & 25.54 d & mon & 512 s & 3996 + 0221640090 & nsvs 12485452&sra02&oct 11 , 2008 & nov . 12 , 2008 & 31.75 d & chr & 32 s & 73326 + [ runs ] to solve the ambiguous spectral classification of corot 0102618121 and to validate our classification criteria , we obtained three high - resolution spectra on 2012 january 10 and 11 with the harps spectrograph installed at the 3.6-m telescope of the eso - la silla observatory . the spectra were taken in the high - efficiency eggs configuration ( r=80,000 ) and they have signal - to - noise ratios ( s / n ) between 50 and 55 . we reduced and normalized the spectra with a semi - automated pipeline developed at the brera observatory @xcite . we then computed the mean line profiles for each spectrum using the lsd software @xcite and we used them to estimate the radial ( from 47.9 to 53.6 kms@xmath13 ) and projected rotational ( from 9.64 to 11.85 kms@xmath13 ) velocities by fitting the profiles with gaussians . the reduced and normalized spectra were shifted by their radial velocities and averaged in order to lessen the effect of the pulsation . the spectral synthesis software sme version 4.2.3 @xcite was used to estimate the physical parameters . the spectral regions 5160 - 5190 ( including the gravity sensitive mg - triplet ) and 4830 - 4900 ( the h@xmath14 region , sensitive to the temperature ) were selected . we left temperature ( @xmath15 ) , gravity ( @xmath16 , cgs units ) , metallicity ( [ fe / h ] ) , micro- and macroturbulence as free parameters , while @xmath17kms@xmath13 was imposed . following @xcite , we also left the abundances of na , si , ti , fe , and ni as free parameters . we performed an iterative process on the marcs stellar models @xcite : we started running the sme software on the mg region , then we switched to the h@xmath14 region keeping the @xmath16 previously found as a fixed parameter . after that we went back to the mg region keeping fixed the @xmath15 value found in the h@xmath14 region , and so on until the results converged . the resulting parameters are @xmath15= 6087@xmath344k , @[email protected] dex , [ fe / h]=@xmath18 dex . they confirm that corot 0102618121 is a pop . i supergiant and not a main - sequence star : its classification as a cepheid variable was definitely established . the original timeseries of corot 0102618121 consist of 348 702 and 206 528 data points , respectively . in both cases the exposure time was set to 32 s ( corresponding to the corot short cadence ) and the flux was recorded in three different spectral regions , i.e. , in the corot chromatic mode . the contamination factors ( in a scale ranging from 0 to 1 ) are 0.026 and 0.036 in the two corot runs , suggesting no relevant contribution from back- or fore - ground stars . the analysis of the original n2 data clearly detected a predominant peak at @xmath19=0.3574 d@xmath13 followed by several harmonics . figure [ lruns ] shows lra01 ( top panel of the upper part ) and lra06 ( top panel of the lower part ) corot n2 data on a 15-d time baseline . all data points are plotted , including those acquired during the crossing of the south atlantic anomaly ( saa ) : they appear as bright outliers on a vertical line . we avoided any cleaning procedure since the detection of instrumental trends and jumps and their subsequent correction is a crucial step to proceed in the analysis of the corot data . therefore , we computed a preliminary solution based on the set of frequencies @xmath19 , 2@xmath19 , 3@xmath19 , 4@xmath19 , and 5@xmath19 . the plots of the residuals ( middle panels of the two parts of fig . [ lruns ] ) show sudden jumps around jd 2454468 in lra01 and jd 2456003 in lra06 , the saa outliers , and continuous oscillations . outliers were iteratively removed at this stage . we still remained with the necessity to correct the sudden jumps and long - term drifts and to verify the origin of the residual oscillations . to do this , we applied the least - squares fits to each pulsation cycle , still using the set of frequencies from @xmath19 to 5@xmath19 . the 32-s cadence of the corot data warranted more than 7000 measurements for each cycle , making the solution very robust . this procedure returned the mean fluxes of the 47 cycles observed in the lra01 and the 27 ones in lra06 ( fig . [ lruns ] , bottom panels of the two parts ) . these mean values show the variations due to the jumps and the long - term trend . we notice that the long - term trends behave in different ways in the two runs . the effect of the jump around jd 2454468 was noticeable in the data until jd 2454485 , thus affecting the mean magnitudes of the subsequent six cycles . to remove the instrumental variations , we interpolated the values of the mean fluxes ( bottom panels of fig.[lruns ] ) at the times of each individual measurement and then we subtracted it . at this point two cleaned timeseries ( 326,634 and 196,032 points , respectively ) were ready to be analysed in frequency . by applying this procedure we probably modified any possible additional signal around 2.8 d and this fact has to be carefully considered when discussing the frequency content of the light curves of the two runs . we transformed the fluxes into magnitudes by calculating the average flux of each run . we also measured the level of the noise of the corot timeseries after subtracting the set of the @xmath19 frequency and harmonics . in the case of the lra01 data we obtained 0.065 mmag in the 0.01 - 1.0 d@xmath13 range , rapidly decreasing to 0.018 mmag in the 1.0 - 2.0 d@xmath13 region , and more slowly to 0.012 mmag in the 2.1 - 3.0 d@xmath13 region and to 0.007 mmag around 5.0 d@xmath13 . in the case of the lra06 data the four values are 0.104 , 0.030 , 0.020 , and 0.010 mmag , respectively . the number of measurements ( 326,634 versus 196,032 ) accounts for most of the difference between the noise levels of the lra01 and lra06 spectra . the frequency analysis was performed by means of the iterative sine - wave , least - squares fitting method @xcite and then refined with the mtrap algorithm @xcite , allowing us to include all the significant harmonics in the search for the best fit around @xmath19 . the best way to illustrate our final results is to follow the chronological order of their progressive consolidation . the analysis of the lra01 data in white light was concluded in 2011 and the results were apparently very promising . we found the harmonics up to 6@xmath19 ( amplitude 0.13 mmag ) , flanked by some modulation terms at @xmath30.007 d@xmath13 ( i.e. , 142 d ) . since the separation is very close to the length of the observations , the origin of these terms has to be ascribed to a residual , still imperfect correction of the instrumental drifts . moreover , a few peaks not very close to @xmath19 and 2@xmath19 were detected : 0.488 d@xmath13 ( amplitude 0.40 mmag ) , 0.392 d@xmath13 ( 0.14 mmag ) , and 0.288 d@xmath13 ( 0.15 mmag ) . only the peak at 0.488 d@xmath13 had a s / n above the threshold considered as the significativity one in the fourier analysis ( s / n=4.0 ; * ? ? ? * ) , i.e. , s / n=0.396/0.065=6.1 . the power spectra illustrating the bunch of low amplitude peaks still centred around @xmath19 , the 0.488-d@xmath13 peak , and the 6@xmath19 harmonic are shown in the top panel of fig . [ spectrares ] . however , the identification of the 0.488 d@xmath13 peak as an additional mode was not really convincing because no combination term was observed . why this non - detection is so important ? the combination or coupling terms are generally ascribed to the non - linear responses of the stellar atmospheres to the strong radial pulsations . the frequency analysis of the light curves of double - mode cepheids show several of them , also involving harmonics of the @xmath1 and @xmath2 radial modes @xcite . the fourier decomposition of the light curves also shows regularities in the distribution of the phase parameters that are not yet fully understood @xcite . the combination terms have been detected in the light curves of other multimode pulsators , e.g. hads stars @xcite , rr lyr stars @xcite , and white dwarfs @xcite . the corot photometry also allowed us to detect low - amplitude , nonradial modes in an hads star pulsating in a single radial mode and also in this case a lot of combination terms were detected @xcite . therefore , we should expect combination terms when additional modes , both radial and nonradial , are excited in a high - amplitude radial pulsator . the uncertainties on the real significativity of the peaks detected in the lra01 data slowed down our theoretical analysis and made us unsure of how to proceed further . the re - observation of the same field in the lra06 pointing supplied us with new data to verify the frequency content of corot 0102618121 . we applied the same procedure used for the lra01 data to remove the content of the main variation . the frequency analysis of the residuals of the lra06 data confirmed the presence of the 0.488 d@xmath13 peak ( fig . [ spectrares ] , bottom panel of the upper part ) . therefore , the confidence on the stellar origin of the 0.488 d@xmath13 grew . however , the ratio 0.357/0.488=0.732 would be quite unusual for a double - mode cepheid with a normal metallic content , both from observational data @xcite and theoretical models @xcite . therefore , such a ratio pointed out a possible nonradial mode . the identification of the peak as the rotational frequency is much more unlikely , since the corresponding rotational period ( 2.06 d ) appears too short for a supergiant star . the excitation of nonradial modes in a classical cepheid would be a very important ingredient in the modelling of the atmospheres of these supergiant stars , thus we submitted this result to further checks . at this stage , we detected the 0.488 d@xmath13 peak in the residuals of each long run . these sets of residuals were obtained by applying a global fit to the lra01 data ( grey points in the upper panel of fig . [ curveres ] ) and to the lra06 ones ( grey points in the lower panel of fig . [ curveres ] ) . indeed , variations in the shape of the residuals are clearly visible , especially after the jumps at jd 2454468 and jd 2456003 . they remind the oscillations noticed before removing the long - term trends and the jumps ( fig . [ lruns ] , middle panels ) . this suggested that the computation of the mean fluxes was very effective in cleaning the power spectra , but the interplay betweeen the large amplitude of corot 0102618121 and the jump correction could somehow leave spurious variability close to the significance level . for this reason we gave a closer look to the light curves of the residuals from which the 0.488 d@xmath13 peak was detected . these light curves show the same type of oscillations , suggesting that the global solution was not able to match the instrumental alterations of the stellar signal . thus we decided to calculate residuals following the same approach used to determine the mean fluxes . this implied to compute as many sets of parameters as the elapsed cycles were , i.e. , 47 for lra01 and 27 for lra06 . the analysis of these new sets of residuals were enlightening . the oscillations visible in fig . [ lruns ] ( middle panels ) had completely disappeared and the light curves became really flat , with just some clues of high - frequency noise ( fig . [ curveres ] , black points ) . indeed , the power spectra of the residuals from fitting one cycle at a time show the peak at the 6@xmath19 harmonic only . this was observed both in the lra01 data and in the lra06 data ( lower part of fig . [ spectrares ] ) . both spectra are characterized by an almost white noise . it could be argued that the signal at 0.488 d@xmath13 was removed when fitting the single cycles ( also taking into account the almost exact 3/4 ratio between the frequencies ) . however , the fact that the 6@xmath19 peak has remained clearly visible gave evidence that the fit with @xmath19 , 2@xmath19 , 3@xmath19 , 4@xmath19 , and 5@xmath19 of each cycle did not modified any signal component apart from these frequencies . we tried to come back to a new correction of the mean fluxes , but again the analysis of the global residuals thus obtained was unsatisfactory . we note that the technique of analysing single cycles were successfully applied to remove similar instrumental effects in the light curves of the high - amplitude @xmath12 sct star corot 101155310 @xcite and of the triple - mode cepheid corot 0223989566 @xcite . l llll c llll & & & + & & & & & & & & & + @xmath20[rad ] & 4.1101(7 ) & 4.220(2 ) & 4.070(2 ) & 3.934(2 ) & & 4.1081(9 ) & 4.191(2 ) & 4.012(6 ) & 3.925(6 ) + @xmath21[rad ] & 4.645(6 ) & 4.637(13 ) & 4.499(21 ) & 4.683(20 ) & & 4.689(7 ) & 4.674(15 ) & 4.814(84 ) & 4.920(64 ) + @xmath22 & 0.05395(4 ) & 0.05371(9 ) & 0.05260(12 ) & 0.05166(11 ) & & 0.05646(5 ) & 0.05717(14 ) & 0.05265(34 ) & 0.04961(30 ) + @xmath23[rad ] & 3.624(7 ) & 3.857(18 ) & 3.431(21 ) & 3.240(17 ) & & 3.573(7 ) & 3.712(26 ) & 3.290(34 ) & 3.193(32 ) + full ampl . [ mag ] & 0.222 & 0.191 & 0.250 & 0.306 & & 0.233 & 0.208 & 0.288 & 0.364 + res . rms [ mag ] & 0.0017 & 0.0036 & 0.0061 & 0.0063 & & 0.0018 & 0.0045 & 0.0150 & 0.0165 + [ colori ] once that we could rule out the excitation of other modes , we fitted the corot magnitudes by means of a cosine series ( see appendix [ appen ] ) . table [ sol ] lists the coefficients of the solutions of the light curves of both runs . the shape of the light curve could be defined by the fourier parameters ( see appendix [ appen ] ) and table [ colori ] lists those relevant to our discussion . the absence of any relevant peak once that the contribution of the main oscillation has been removed is not by itself an evidence that the light curve is stable . cycle - to - cycle variations as those observed in the _ kepler _ light curve of v1154 cyg or in the radial velocity curves could not result in a well defined peak in the power spectrum . moreover , at least in principle , the residual peaks left around @xmath19 when applying a global fit could be ascribed to cycle - to - cycle variations , maybe erratic ones . we analysed the @xmath22 , @xmath20 , and @xmath21 parameters of each cycle to quantify how much the light curve deviates from the average behaviour . the corot 32-s cadence allowed us to get about 7500 measurements per cycle , adequate to survey the harmonics 2@xmath19 and 3@xmath19 in a very accurate way , though their amplitudes are very small ( 6.0 and 0.7 mmag , respectively ) . the fourier parameters are shown in fig . [ fourier ] . the error bars are three times the standard deviations ( @xmath0 ) around the average of the parameters of the individual cycles . the immediate result is that the differences from the average values are small . we could notice that all the individual values are between 3@xmath24 , though some apparently regular variations are sometimes visible ( e.g. the @xmath22 behaviour in the lra01 ) . therefore , we performed a frequency analysis of the fourier parameters . we could not detect any periodicity since the power spectra of the @xmath22 , @xmath20 , and @xmath21 parameters do not show any common feature either in a single long run or when comparing lra01 and lra06 results . however , the best parameter to quantify the stability of the light curve is probably the o - c parameter . we determined the observed t@xmath5 s by means of cubic spline functions with a non - zero smoothing parameter that depends on the number of measurements to fit and on their scatter @xcite . we thus obtained the linear ephemeris from the least - squares method @xmath25 + ( e@xmath26 ) and hence we could calculate the o - c value for each t@xmath5 . the residual rms of the least - squares linear fit is 0.0052 d and the 3@xmath24 band is shown in fig . [ fourier ] . we also estimated the uncertainty on the individual t@xmath5 determination as the difference between the two times corresponding to the intersection of the spline function with the line @xmath27 ( fig . [ spline ] ) , where @xmath28 is the magnitude at maximum and @xmath24 is the standard deviation of the fit . such uncertainties are ranging from 0.010 to 0.016 d and hence the 3@xmath24 interval ( i.e. , 22.5 min ) matches the individual error bars , too . we analysed in frequency the o - c values too , again obtaining no clear indication of periodicities . as a final conclusion , looking at the different plots shown in fig . [ fourier ] and at the results of the various frequency analyses , we can not identify any additional periodicity in the variability of corot 0102618121 . the fourier parameters and the o - c values sometimes mimic an apparent regular behaviour . this could be a hint of cycle - to - cycle variations , much less regular than those observed in the _ kepler _ light curve of v1154 cyg @xcite . we merged the lra01 and lra06 measurements and we calculated a refined value of the period on the whole time baseline . the mtrap code supplied the final vale @[email protected] d , in excellent agreement with the value obtained from the t@xmath5 s values only . the most relevant difference in the datasets of the two long runs is a slight increase from 0.222 to 0.233 mag in the full - amplitude of the light curves . since we are dealing with absolute photometry , i.e. , without any comparison star , we have to be sure that the measurements were performed in the same instrumental system to infer that the amplitude difference has a stellar origin . we immediately noticed that the star was observed in the same ccd in the two runs , but with a different mask and at different pixel coordinates . as a results , the average white flux measured was of 1 030 069 electrons in lra01 and 944,911 electrons in lra06 . moreover , the flux subdvision in @xmath30 , @xmath31 , @xmath32 colours was also different : 65 , 14 , 21 per cent in lra01 , 78 , 10 , 12 per cent in lra06 . we conclude that the small difference in amplitude could very probably be ascribed to the different settings of the corot pointings in the two runs . the same effect occurs in _ kepler _ photometry : average fluxes and amplitudes change for the same star observed in the same ccd but with different masks . these effects are particularly disturbing when dealing with large amplitude variables @xcite . the procedure applied to white light measurements was repeated for each colour of both runs . other similar , subtle instrumental effects became evident during this process @xcite . then , the fourier decomposition was applied to the cleaned timeseries in each colour ( table [ colori ] ) . the parameters clearly show the expected dependence from the bandpass ( see * ? ? ? * for a comparison between @xmath33 and @xmath34 parameters ) . the different flux distributions between lra01 and lra06 also account for the small differences between parameters in the same colour . the procedure described in the previous sections for corot 0102618121 was applied to the other cepheids , whose folded light curves are shown in fig . [ curve ] ( six pop . i stars ) and fig . [ curvaii ] ( one pop . ii star ) . the fourier parameters are listed in table [ others ] . the frequency analysis of the v376 mon data supplied @xmath19=0.6060 d@xmath13 , refined to @xmath35 d with mtrap . table [ sol ] lists the least - squares solution of the corot white light curve . the frequency analysis did not detect any significant additional mode , but again the residual power spectra strongly depend on the procedure used to prewhiten the data . we obtained a bunch of peaks around @xmath19 when applying a global fit up to 7@xmath19 to the entire dataset . these peaks disappeared when the fit was calculated cycle - by - cycle ( fig . [ cef751 ] , upper part ) . the harmonic 8@xmath19 was detected in both cases , confirming that the single cycle fit did not alter the frequency content . the fourier parameters and the o - c values show a small scatter around the means ( fig . [ decocef751 ] ) , suggesting no periodic variation . the frequency analysis of corot 0659466327 data returned @xmath19=0.648 d@xmath13 and the period was refined to @xmath36 d. table [ solii ] lists the least - squares solution of the white light curve . the bump on the ascending branch ( fig . [ curvaii ] ) is typical of pop . ii cepheids . differently than in the previous cases , the peak in the power spectrum of the residuals from the global fit up to 9@xmath19 occurred around 2@xmath19 , but the bunch of peaks at @xmath19 were detected , too . the power spectrum of the residuals from the cycle - by - cycle fits is completely flat and it just shows the 10@xmath19 harmonic , not prewhitened from the data ( fig . [ cef751 ] , lower part ) . small - size scale variations are observed in the fourier parameters and o - c values , but again without any trace of periodicity ( fig . [ decocef327 ] ) . the power spectra of the residuals of corot 0102618121 , v376 mon and corot 0659466327 calculated by means of a global least - squares solution show the limits of this approach when applied to continuous and high - precision timeseries where systematic errors ( jumps and/or drifts ) can occur . when possible , e.g. in presence of a high - amplitude predominant mode , the frequency analysis of cycle - to - cycle residuals are preferable , taking care to keep a known frequency , e.g. an harmonic , as internal check of the signal preservation . the solutions of the light curves of the cepheids corot 0221640090 ( table [ soliii ] ) , corot 0211626074 ( table [ soliv ] , corot 0315221415 ( table [ solv ] ) , and dp mon ( table [ solvi ] were much less laborious since these stars were observed once in some short runs ( srs ) and hence followed for a restricted number of cycles . the o - c plots do not provide useful hints about periodic variations ( fig . [ ocallcep ] ) . only in the case of corot 0221640090 ( bottom panel ) the string of consecutive t@xmath5 shows a hint of regular behaviour , but the monitoring is too short to assess its repetitivity on a long time - scale . moreover , the standard deviation around the linear ephemeris is @xmath30.0033 d and all the maxima are well within 3@xmath24 . ll lll llll & & & & & & & & + & & & & & & & & + v376 mon & @xmath1 & 4785.4731 & 1.651969(1 ) & 0.372 & 4.1811(3 ) & 0.3745(1 ) & 2.2206 ( 1 ) & 0.172(1 ) + corot 0221640090 & @xmath2 & 4754.6755 & 2.19758(4 ) & 0.288 & 4.560(1 ) & 0.1397(1 ) & 3.641(5 ) & 2.670(9 ) + dp mon & @xmath1 & 4754.0407 & 3.08704(4 ) & 0.678 & 4.312(1 ) & 0.5201(1 ) & 2.4703(4 ) & 0.5439(6 ) + corot 0315221415 & @xmath1 & 5258.4566 & 4.7085(17 ) & 0.557 & 4.363(1 ) & 0.4650(5 ) & 2.478(3 ) & 0.434(5 ) + corot 0211626074 & @xmath1 & 4199.7103 & 5.4706(2 ) & 0.296 & 4.731(1 ) & 0.3147(3 ) & 3.242(3 ) & 1.391(35 ) + corot 0659466327 & pop . ii & 6031.4482 & 1.542353(2 ) & 0.492 & 4.394(1 ) & 0.2913(2 ) & 0.886(1 ) & 4.953(2 ) + the fourier decomposition of the light curves of cepheids allowed us to disentangle @xmath1 and @xmath2 radial pulsators ( fig . [ grafo ] ) . the @xmath20-sequence formed by the classical cepheids reproduces the well - known hertzsprung progression . it can be described by the linear relation @xmath37 @xcite . the progression is sharp , since a shift of @xmath30.30 rad pushes a star well outside it . another z - shaped sequence was put in evidence by applying the fourier decomposition to the light curves of galactic cepheids with @xmath388 d : the upper part of this sequence is composed of stars with @[email protected] rad , the lower part of stars with @xmath20@xmath40 4.0 rad ( fig . [ grafo ] , top panel ) . the @xmath21@xmath41 plane was the rosetta stone to link the upper and lower sequences , since it shows just one continuous sequence , well separated from the classical one ( fig . [ grafo ] , middle panel ) . the fourier decomposition of the light curves of double - mode ( @xmath42 ) cepheids provided the definitive evidence that the pulsation mode drives the separation between the different sequences @xcite . the @xmath22@xmath41 plot shows that the amplitude of the harmonic 2@xmath19 of the @xmath2-mode pulsators is much smaller than that of the main frequency @xmath19 ( fig . [ grafo ] , bottom panel ) . this fact makes the light curves of the @xmath2-mode pulsators much more ( but not exactly ) sine - shaped than those of @xmath1-mode pulsators . theoretical works demonstrated how the inclusion of convective energy transport has made possible to reproduce the observed sequence of @xmath2-pulsators and that the sharp features in the fourier coefficients are due to the resonance @xmath43 between the fourth and first overtones @xcite . the classification of corot 0211626074 and corot 0315221415 is straightforward , since the @xmath20 , @xmath22 , and @xmath21 parameters put both these long - period ( in our sample ) cepheids on the well - defined sequence of @xmath1-mode pulsators . the period of corot 0102618121 and dp mon are very particular in this context . they lie exactly on the discontinuity of the z - shaped sequence of the @xmath2 pulsators in the @xmath20@xmath41 plane , where the intersection with the @xmath1-sequence occurs . therefore , the @xmath20 parameter alone is not able to identify their pulsation mode . however , the much higher value of @xmath21 and the much smaller value of @xmath22 with respect to those of the classical cepheids clearly demonstrate that corot 0102618121 is a @xmath2 pulsator . on the other hand , the @xmath21 and @xmath22 parameters of dp mon match the @xmath1-sequence . the light curve of dp mon is very asymmetrical ( fig . [ curve ] ) and it is very similar to those of rrab stars . in particular , notice the kink before the minimum , that requires a large number of harmonics to be fitted . we classify corot 0221640090 as a @xmath2 pulsator , like corot 0102618121 . the case of v376 mon is more intriguing , since its period is very short also with respect to the full sample of galactic cepheids . the cepheids with @xmath382.5 d are quite rare and the space of fourier parameters is scarcely populated in this region . for this reason the values of @xmath20 and @xmath21 are not able to provide a clear mode identification . however , the @xmath22 value is quite high , and hence the star is very probably a @xmath44mode pulsator . the continuous photometry allowed us to verify the phases of maximum brightness ( @xmath45 ) in the corot colours with great accuracy . the analyses described in the previous sections were repeated for each colour of each star . the results are summarized in table [ coloriall ] . there is a common trend indicating that the maximum brightness occurs a little earlier at short wavelengths than at long ones . the only exception is dp mon , whose maxima are measured at the same phases ( within error bars ) in all colours . the observed phase shifts are very small , around 0.01 period . this result strongly corroborates the practice to merge all the maxima observed with different filters or techniques when studying secular period variations . indeed , this shift is usually smaller than the uncertainties on the t@xmath5 determination in case of ground - based , fragmented observations . this is the first measurement of the phase shifts among light curves of cepheids in different colours after the pioneering approach by @xcite . l rrrr & & & & + corot 0102618121 & 0.993 & 0.997 & 1.004 & 1.000 + in lra01 & @xmath30.003 & @xmath30.003 & @xmath30.003 & @xmath30.003 + corot 0102618121 & 0.994 & 0.996 & 1.003 & 1.000 + in lra06 & @xmath30.005 & @xmath30.005 & @xmath30.003 & @xmath30.003 + corot 0221640090 & 0.994 & 1.002 & 1.007 & 1.000 + & @xmath30.003 & @xmath30.003 & @xmath30.005 & @xmath30.003 + v376 mon & 0.996 & 0.998 & 1.002 & 1.000 + & @xmath30.003 & @xmath30.003 & @xmath30.003 & @xmath30.003 + dp mon & 0.998 & 1.000 & 1.001 & 1.000 + & @xmath30.003 & @xmath30.003 & @xmath30.003 & @xmath30.003 + the power spectra of the corot light curves of galactic cepheids did not show any convincing evidence of the excitation of nonradial modes , especially in the three monitored in the long runs . stars showing additional modes were found with an incidence of 10% among the @xmath2-cepheids of the lmc @xcite . the only two @xmath2-pulsators observed by corot can not yet tell us if galactic @xmath2-cepheids behave differently than lmc ones . on the other hand , the @xmath1-cepheids of the lmc do not show additional modes @xcite and the intensive corot photometry does not modify this issue . the detection of the same peak in two independent sets of residuals of corot 0102618121 seemed to be a decisive piece of evidence in favour of the excitation of a nonradial mode , but it not survived the change of the procedure for the computation of the residuals . in our opinion the instrumental effects seem to play an important role in the long time series of space photometry . our recommendation is to analyse residuals calculated from local solutions ( e.g. cycle - to - cycle when possible ) to search for additional periodicities . instrumental drifts and jumps ( fig . [ lruns ] ) or , in extreme cases , numerical errors or amplitude changes can enhance spurious peaks in the power spectrum of the residuals calculated from a global fit ( fig . [ spectrares ] ) . the residuals from the cycle - by - cycle fits are providing a much more reliable dataset to be analysed in frequency ( fig . [ curveres ] ) . omitting a high harmonic from the fit and verify that it is correctly recovered is an excellent litmus test of the whole procedure . moreover , the lack of combination terms between the main pulsation mode and the suspected additional component should be considered a warning against the identification of this component as a true excited mode . another compelling result expected from the analysis of the corot light curves was to verify the repetitivity and the amplitude of the cycle - to - cycle variations , as those observed in the _ kepler _ data of v1154 cyg @xcite and in the radial - velocity surveys @xcite . the fourier parameters of the light curves of corot cepheids show some variability , but in general all the changes are within @xmath0 . moreover , they seem to be erratic , not reproducing phenomena like activity cycle or long - term effects . no large scatter was observed around the maxima of the folded light curves and no large variations were observed in the o - c plots of the cepheids monitored with corot . in particular , the o - c values are all within @xmath30.01 d , i.e.,@xmath315 min , including the @xmath2-mode pulsators corot 0102618121 and corot 0221640090 ( figs . [ fourier ] , [ decocef751 ] , [ decocef327 ] , and [ ocallcep ] ) . this is particularly interesting since the space photometry obtained with the satellite _ most _ showed some instabilities in the pulsation cycle of the @xmath2-cepheid sz tau , while the @xmath1-mode rt aur repeated the light curve more precisely @xcite . therefore , we performed some additional tests on the folded light curves , evaluating the standard deviations as a function of phase bins . tiny increases of the standard deviations were measured in correspondence of the most rapid light variation ( i.e. , on the steep ascending branch ) ; they mostly reflect the variability due to the pulsation within the phase bins , as proven by the fact that they appear to be more relevant for the stars showing the largest amplitudes and the most asymmetrical light curves . this suggests that the standard deviations of the phase bins on the rising branch can not be used as a tool to search for cycle - to - cycle variations . we can conclude that in the case of the seven corot cepheids , the cycle - to - cycle variability is very limited and not greatly affecting the repetitivity of the light curves . the light variations observed around the maxima of sz tau are more promising , but on this context the low - frequency peaks in the power spectrum have to be carefully considered : the residuals observed in _ most _ photometry are very similar to the spurious ones detected in corot one ( fig . [ curveres ] ) . we also benefit from the unique opportunity provided by the corot multicolour photometry . very small shifts in the t@xmath5 s were observed for the first time in the optical range . because these shifts are around 0.01 period only , our analysis support the use of t@xmath5 s observed in different passbands to reconstruct secular period variations ( e.g. * ? ? ? * ) . the accurate values of the fourier parameters provided by the available space photometry up to very high harmonics can be used to improve the automatic classification of the cepheid subclasses , i.e. , pop . i , pop . ii , @xmath1-mode , and @xmath2-mode pulsators . the quantitative separation of these subclasses in the parameter space is of particular relevance when considering the large harvest of new variables that will be discovered by future , wide - field , and all - sky space missions like tess @xcite and plato 2.0 @xcite . the corot space mission has been developed and operated by cnes , with contributions from austria , belgium , brazil , esa ( rssd and science program ) , germany , and spain . the authors thank the anonymous referee for useful comments . ep acknowledges observatoire midi - pyrnes for the two - months grant allocated between 2014 may and july ; useful discussions with boris dintrans and pascal fouqu at the very beginning of the project are acknowledged . jmb thanks for the support of the nkfih grant k-115709 . mr acknowledges financial support from the fp7 project _ spaceinn : exploitation of space data for innovative helio- and asteroseismology _ , ep acknowledges prin - inaf 2014 _ galactic archaelogy _ for support to latest activities . this research has made use of the exodat database , operated at lam - oamp , marseille , france , on behalf of the corot / exoplanet program . the present study has used the simbad data base operated at the centre de donnes astronomiques ( strasbourg , france ) . mnras anderson r.i . , 2014 , , 566 , l10 antonello e. , poretti e. , reduzzi l. , 1990 , , 236 , 138 benk j.m . , plachy e. , szab r. , molnr l. , kollth z. , 2014 , , 213 , 31 berdnikov l.n . , kniazev a.yu . , sefako r. , kravtsov v.v . , zhujko s.v . , 2014 , astronomy letter , 40 , 125 bruntt h. , et al . , 2008 , , 683 , 433 buchler j.r . , 2008 , , 680 , 1412 buchler j.r , szab r. , 2007 , , 660 , 723 burki g. , mayor m. , benz w. , 1982 , , 109 , 258 carpino m. , milani a. , nobili a.m. , 1987 , a&a , 181 , 182 debosscher j. , et al . , 2009 , , 506 , 519 deleuil m. , et al . , 2009 , , 138 , 649 derekas a. , et al . , 2012 , , 425 , 1312 donati j .- f . , semel m. , carter b.d . , rees d.e . , collier cameron a. , 1997 , , 291 , 658 evans n.r . , sasselov d.d . , short i.s . , 2002 , , 567 , 1121 evans n.r . , et al . , 2015 , , 446 , 4008 feuchtinger m. , buchler j.r . , kollth z. , 2000 , , 544 , 1056 fruth t. , et al . , 2012 , , 143 , 140 gruberbauer m. , et al . , 2007 , , 379 , 1498 gustafsson b. , edvardsson b. , eriksson k. , jrgensen u.g . , nordlund . , plez b. , 2008 , , 486 , 951 kabath p. , eigmller p. , erikson a. , hedelt p. , rauer h. , titz r. , wiese t. , karoff c. , 2007 , , 134 , 1560 kabath p. , et al . , 2008 , , 136 , 654 kabath p. , et al . , 2009 , , 137 , 3911 kabath p. , et al . , 2009 , , 506 , 569 kanev e. , savanov i. , sachkov m. , 2015 , in 3@xmath46 corot symposium , 7@xmath7 kasc meeting , the space photometry revolution , toulouse , july 2014 , in press karoff c. , et al . , 2007 , , 134 , 766 kuschnig r. , weiss w.w . , gruber r. , bely p.y . , jenkner h. , 1997 , , 328 , 544 molnr l. , szabados l. , 2014 , , 442 , 3222 montgomery m.h . , odonoghue d. , 1999 , delta scuti star newsletter , 13 , 28 morgan s.m . , simet m. , bargenquast , s. , 1998 , , 48 , 509 moskalik p. , 2014 , in guzik j.a . , chaplin w.j . , handler g. , pigulski a. , eds . , iau proceedings s301 , vol . 9 , precision asteroseismology , p. 249 moskalik p. , koaczkowski z. , 2008 , comm . in asteroseismology , 157 , 343 moskalik p. , koaczkowski z. , 2009 , , 394 , 1649 neilson h.r . , ignace r. , 2014 , , 563 , l4 pardo i. , poretti e. , 1997 , , 324 , 121 poretti e. , 2000 , in ibanoglu c. , ed . , nato science series , vol . 544 , variable stars as essential astrophysical tools , kluwer academic publishers , p. 421 poretti e. , baglin a. , weiss w.w . , 2014 , , 795 , l36 poretti e. , le borgne j.f . , klotz a. , 2015 , in 3@xmath46 corot symposium , 7@xmath7 kasc meeting , the space photometry revolution , toulouse , july 2014 , preprint ( arxiv:1506.08122 ) poretti e. & pardo i. , 1997 , , 324 , 133 poretti e. , et al . , 2011 , , 528 , a147 poretti e. , et al , 2005 , , 440 , 1097 rainer m. , 2003 , laurea thesis , universit degli studi di milano rauer h. , et al . , 2014 , exp . astron . , 38 , 249 reinsch h. , 1967 , numerische mathematik 10 , 177 ricker g.r . , et al . , 2014 , spie proc . , 9143 , 20 spreckley s.a . , stevens i.r . , 2008 , , 388 , 1239 szildi k. , vink j. , poretti e. , szabados l. , kun m. , 2007 , , 473 , 579 valenti j.a . , fischer d.a . , 2005 , , 159 141 valenti j.a . , piskunov n. , 1996 , , 118 , 595 vaniek p. , 1971 , , 12 , 10 wachmann a.a . , 1964 , astron . abhandlungen hamburger sternwarte , 7 , 4 wisniewski w.z . , johnson h.l . , 1968 , communications of the lunar and planetary laboratory , volume 7 , part 2 , p. 57 - 78 wu y. , 2001 , , 323 , 248 we provide the solutions of the corot light curves of the cepheids observed in the long runs . we fitted the corot magnitudes by means of the formula @xmath47}\ ] ] and we calculated the fourier parameters @xmath48 ( e.g. @xmath49 ) and @xmath50 ( e.g. @xmath51 and @xmath52 ) from the amplitude ( @xmath53 ) and phase ( @xmath54 ) coefficients . the error bars were calculated at the @xmath55 level @xcite . r r ll c ll c ll & & & + & & & & & + & & & & & & & & & + & & & & & & & & & + @xmath19&&0.110500(4 ) & 3.06375(4 ) & & 0.116054(6 ) & 3.05962(5 ) & & 0.15852(2 ) & 2.5446(1 ) + 2@xmath19&&0.005962(4 ) & 3.95440(72 ) & & 0.006552(6 ) & 3.94414(90 ) & & 0.05937(2 ) & 2.9871(2 ) + 3@xmath19&&0.000699(4 ) & 1.26965(616 ) & & 0.000896(6 ) & 1.30168(655 ) & & 0.02715(2 ) & 3.5712(5 ) + 4@xmath19&&0.000621(4 ) & 3.31223(693 ) & & 0.000809(6 ) & 3.24492(726 ) & & 0.01096(2 ) & 4.0676(14 ) + 5@xmath19&&0.000287(4 ) & 4.55196(1499 ) & & 0.000389(6 ) & 4.51949(1509 ) & & 0.00435(2 ) & 4.4503(34 ) + 6@xmath19&&0.000135(4 ) & 5.74428(3180 ) & & 0.000120(6 ) & 5.64753(4876 ) & & 0.00156(2 ) & 4.6689(94 ) + 7@xmath19&&0.000049(4 ) & 0.55276(8822 ) & & 0.000078(6 ) & 0.09852(7486 ) & & 0.00077(2 ) & 4.5913(190 ) + 8@xmath19&&0.000024(4 ) & 2.10979(17710 ) & & 0.000025(6 ) & 1.97647(23114 ) & & 0.00049(2 ) & 4.6907(297 ) + 9@xmath19 & & & & & & & & 0.00036(2 ) & 4.9298(408 ) + 10@xmath19 & & & & & & & & 0.00028(2 ) & 5.3570(534 ) + 11@xmath19 & & & & & & & & 0.00016(2 ) & 5.8615(952 ) + 12@xmath19 & & & & & & & & 0.00010(2 ) & 6.1633(1511 ) + 13@xmath19 & & & & & & & & 0.00007(2 ) & 0.7590(2155 ) + 14@xmath19 & & & & & & & & 0.00005(2 ) & 0.7422(3075 ) + & & & & & + & & & & & + [ sol ] r r ll & + & + & & & + & & & + @xmath19 & & 0.19348(4 ) & 2.6323(2 ) + 2@xmath19 & & 0.05637(4 ) & 3.3752(7 ) + 3@xmath19 & & 0.03478(4 ) & 2.5001(12 ) + 4@xmath19 & & 0.02705(4 ) & 2.9155(16 ) + 5@xmath19 & & 0.01608(4 ) & 3.8907(26 ) + 6@xmath19 & & 0.01154(4 ) & 4.7951(36 ) + 7@xmath19 & & 0.00629(4 ) & 5.6173(67 ) + 8@xmath19 & & 0.00262(4 ) & 0.3086(161 ) + 9@xmath19 & & 0.00055(4 ) & 1.1627(767 ) + 10@xmath19 & & 0.00052(4 ) & 6.0158(832 ) + 11@xmath19 & & 0.00076(4 ) & 0.8167(555 ) + 12@xmath19 & & 0.00083(4 ) & 1.9510(508 ) + 13@xmath19 & & 0.00068(4 ) & 2.9567(614 ) + 14@xmath19 & & 0.00041(4 ) & 4.1419(1024 ) + 15@xmath19 & & 0.00019(4 ) & 5.2084(2196 ) + 16@xmath19 & & 0.00007(4 ) & 5.7712(6216 ) + 17@xmath19 & & 0.00009(4 ) & 5.4264(4543 ) + 18@xmath19 & & 0.00016(4 ) & 0.2652(2712 ) + & + & + [ solii ] r r ll & + & + & & & + & & & + @xmath19 & & 0.14079(1 ) & 2.9111(1 ) + 2@xmath19 & & 0.01967(1 ) & 4.0987(7 ) + 3@xmath19 & & 0.00290(1 ) & 6.0910(47 ) + 4@xmath19 & & 0.00156(1 ) & 1.7476(87 ) + 5@xmath19 & & 0.00102(1 ) & 3.1642(132 ) + 6@xmath19 & & 0.00055(1 ) & 4.3818(244 ) + 7@xmath19 & & 0.00035(1 ) & 5.5170(384 ) + 8@xmath19 & & 0.00017(1 ) & 0.4274(795 ) + 9@xmath19 & & 0.00009(1 ) & 1.3975(1468 ) + 10@xmath19 & & 0.00002(1 ) & 1.9638(6112 ) + 11@xmath19 & & 0.00006(1 ) & 3.2910(2157 ) + 12@xmath19 & & 0.00002(1 ) & 3.6209(6119 ) + & + & + [ soliii ] r r ll & + & + & & & + & & & + @xmath19 & & 0.12984(3 ) & 2.5207(3 ) + 2@xmath19 & & 0.04086(3 ) & 3.4897(8 ) + 3@xmath19 & & 0.01028(3 ) & 4.5213(32 ) + 4@xmath19 & & 0.00095(3 ) & 5.1938(341 ) + 5@xmath19 & & 0.00166(3 ) & 3.0263(195 ) + 6@xmath19 & & 0.00128(3 ) & 3.7305(253 ) + 7@xmath19 & & 0.00103(3 ) & 4.5825(315 ) + 8@xmath19 & & 0.00062(3 ) & 5.3574(527 ) + 9@xmath19 & & 0.00036(3 ) & 6.2663(912 ) + 10@xmath19 & & 0.00018(3 ) & 0.8783(1842 ) + 11@xmath19 & & 0.00010(3 ) & 2.1266(3298 ) + 12@xmath19 & & 0.00008(3 ) & 2.7998(4124 ) + & + & + [ soliv ] r r ll & + & + & & & + & & & + @xmath19 & & 0.21116(6 ) & 2.4327(3 ) + 2@xmath19 & & 0.09819(6 ) & 2.9456(6 ) + 3@xmath19 & & 0.04415(6 ) & 3.4927(14 ) + 4@xmath19 & & 0.02011(6 ) & 3.8819(31 ) + 5@xmath19 & & 0.00735(6 ) & 4.0916(84 ) + 6@xmath19 & & 0.00386(6 ) & 3.9842(159 ) + 7@xmath19 & & 0.00298(6 ) & 3.7086(207 ) + 8@xmath19 & & 0.00314(6 ) & 3.8961(196 ) + 9@xmath19 & & 0.00299(6 ) & 4.3491(205 ) + 10@xmath19 & & 0.00246(6 ) & 4.8484(248 ) + 11@xmath19 & & 0.00201(6 ) & 5.4295(305 ) + 12@xmath19 & & 0.00158(6 ) & 6.0745(387 ) + 13@xmath19 & & 0.00122(6 ) & 0.3769(501 ) + 14@xmath19 & & 0.00087(6 ) & 1.1612(706 ) + 15@xmath19 & & 0.00060(6 ) & 1.8127(1016 ) + 16@xmath19 & & 0.00045(6 ) & 2.6720(1363 ) + 17@xmath19 & & 0.00032(6 ) & 3.5748(1915 ) + 18@xmath19 & & 0.00016(6 ) & 4.2500(3960 ) + 19@xmath19 & & 0.00020(6 ) & 5.1683(3062 ) + 20@xmath19 & & 0.00016(6 ) & 0.3755(3767 ) + & + & + [ solv ] r r ll & + & + & & & + & & & + @xmath19 & & 0.23679(2 ) & 2.2927(1 ) + 2@xmath19 & & 0.12315(2 ) & 2.6145(1 ) + 3@xmath19 & & 0.07198(2 ) & 3.0652(2 ) + 4@xmath19 & & 0.04044(2 ) & 3.4315(4 ) + 5@xmath19 & & 0.02352(2 ) & 3.6996(7 ) + 6@xmath19 & & 0.01444(2 ) & 4.0036(12 ) + 7@xmath19 & & 0.00856(2 ) & 4.2711(20 ) + 8@xmath19 & & 0.00523(2 ) & 4.4842(33 ) + 9@xmath19 & & 0.00323(2 ) & 4.6872(54 ) + 10@xmath19 & & 0.00205(2 ) & 4.7897(85 ) + 11@xmath19 & & 0.00126(2 ) & 4.8674(137 ) + 12@xmath19 & & 0.00096(2 ) & 5.0046(180 ) + 13@xmath19 & & 0.00074(2 ) & 5.0513(234 ) + 14@xmath19 & & 0.00058(2 ) & 5.2112(296 ) + 15@xmath19 & & 0.00048(2 ) & 5.3839(359 ) + 16@xmath19 & & 0.00041(2 ) & 5.6321(425 ) + 17@xmath19 & & 0.00034(2 ) & 5.9751(515 ) + 18@xmath19 & & 0.00024(2 ) & 6.0823(712 ) + 19@xmath19 & & 0.00022(2 ) & 0.1627(799 ) + 20@xmath19 & & 0.00014(2 ) & 0.5256(1234 ) + 21@xmath19 & & 0.00016(2 ) & 0.7288(1113 ) + 22@xmath19 & & 0.00010(2 ) & 1.3601(1793 ) + 23@xmath19 & & 0.00009(2 ) & 1.1871(2008 ) + 24@xmath19 & & 0.00003(2 ) & 1.4777(5286 ) + 25@xmath19 & & 0.00005(2 ) & 1.8188(3629 ) + & + & + [ solvi ]	a few galactic classical cepheids were observed in the programmes of space missions as _ coriolis _ , _ most _ and _ kepler_. an appealing opportunity was to detect additional nonradial modes , thus opening the possibility to perform asteroseismic studies and making the pulsational content of galactic cepheids more similar to that of magellanic clouds ones . however , only hints of cycle - to - cycle variations were found , without any strict periodicity . in this context the potential of the corot exoplanetary data base was not fully exploited despite the wide area covered on the galactic plane . therefore , we investigated all the candidate cepheids pointed out by the automatic classification of the corot curves . at the end we could identify seven bona - fide cepheids . the light curves were investigated to remove some instrumental effects . the frequency analysis was particularly delicate since these small effects can be enhanced by the large amplitude , resulting in the presence of significant , but spurious , peaks in the power spectrum . indeed , the careful evaluation of a very attracting peak in the spectra of corot 0102618121 allowed us to certify its spurious origin . once that the instrumental effects were properly removed , no additional mode was detected . on the other hand , cycle - to - cycle variations of the fourier parameters were observed , but very small and always within @xmath0 . among the seven cepheids , there are two pop . i first - overtone pulsators , four pop . i fundamental mode pulsators , and one pop . ii star . the corot colours allowed us to measure that times of maximum brightness occur a little earlier ( about 0.01 period ) at short wavelengths than at long ones . [ firstpage ] methods : data analysis stars : individual : v376 mon stars : individual : dp mon stars : interiors stars : oscillations stars : variables : cepheids
in the last decade , several methods have become available to prepare molecules in the gas phase at low temperatures @xcite . going beyond thermalization with a cold environment , these methods allow for a precise control of the external and internal degrees of freedom in molecular samples . cold molecules with translational temperatures in the range 100mk1k may then find applications in precision spectroscopy @xcite and in the study of chemical reactivity at very low collision energies and very high energy resolution @xcite . alternatively , they may also represent the starting point for further cooling steps toward the ultracold regime below 1mk @xcite . the choice of the cooling method depends on the properties of the species of interest @xcite . in the case of paramagnetic atoms and molecules which can be entrained in a supersonic beam , multistage zeeman deceleration @xcite and related techniques @xcite are the methods of choice : the phase - space density of atoms and molecules in supersonic beams is high and is preserved by a phase - stable operation of the decelerator @xcite , the method can be quantum - state selective @xcite , the final velocity of the particles can be controlled and tuned over a wide range @xcite , and the decelerated particles can be loaded into a magnetic trap @xcite for observation of slow decay processes or spectroscopic measurements requiring long observation times . we describe here the use of multistage zeeman deceleration to generate slow and velocity - tunable beams of he@xmath9 molecules in the metastable @xmath10 state ( referred to as he@xmath11 hereafter ) in a setup tailored for future applications in precision spectroscopy . we also demonstrate the possibility to use the cold molecular samples to study the rovibrational energy - level structure of the @xmath10 state of he@xmath9 and the @xmath12 ground state of he@xmath5 by photoionization and photoelectron spectroscopic methods . being three- and four - electron systems , respectively , he@xmath9 and he@xmath5 represent attractive systems to test the accuracy of _ ab - initio _ quantum - chemical methods . in the two - electron systems h@xmath9 , hd , and d@xmath9 , the results of _ ab - initio _ calculations of dissociation energies and rovibrational energy levels which include adiabatic and nonadiabatic corrections to the born - oppenheimer approximation as well as quantum - electrodynamics corrections up to the leading ( one - loop ) contribution to the term proportional to @xmath13 were found to be in agreement with experimental results within the combined uncertainties of the experimental and theoretical determinations @xcite . such calculations become increasingly challenging as the number of electrons increases , and the magnitudes of the relativistic and quantum - electrodynamics corrections are not accurately known for he@xmath5 . in a recent high - resolution spectroscopic study of the threshold ionization spectrum of he@xmath11 , the level spacings between the lowest four rotational levels of he@xmath5 could be determined at an accuracy of 100mhz @xcite . comparison with the latest _ ab - initio _ calculations neglecting quantum - electrodynamics corrections @xcite did not reveal significant deviations from the experimental results at this level of accuracy , except perhaps for the highest rotational level observed experimentally . in future , we would like to improve the accuracy of these measurements by using a slow beam of metastable he@xmath9 , which motivated us to develop a he@xmath11 source better suited to precision spectroscopic experiments . as a starting point , a supersonic beam of @xmath14 was produced with a pulsed source operated at cryogenic temperatures . the velocity of the molecules was subsequently reduced to @xmath15100 m/s by coupling the cryogenic beam source to a multistage zeeman decelerator . from pulsed - field - ionization zero - kinetic - energy ( pfi - zeke ) photoelectron spectra of the @xmath16 transition , we have determined the internal - state distribution of the cold molecules in the beam and found rotational states up to @xmath2 to be populated . deceleration experiments with these molecules enabled the study of the rotational state selectivity of the deceleration process and a comparison with the deceleration of molecular oxygen ( @xmath17 ) in the @xmath18 state , for which full selectivity of the spin - rotational state @xmath19 was achieved @xcite . @xmath14 is formed in a gas of pure helium under conditions where the density allows for three - body collisions on the relevant time scales and collision partners of sufficiently high energy are available to excite helium atoms to their metastable states . such conditions typically prevail in plasmas , such as those generated by striking an electric discharge through helium gas @xcite . the large ratio of magnetic moment to mass makes @xmath14 particularly attractive for multistage zeeman deceleration . moreover , the predicted radiative lifetime of 18s of @xmath14 @xcite is sufficiently long that no decay takes place on the millisecond timescale of the deceleration process . however , the high velocity ( about 2000 m/s ) of a supersonic expansion of pure he from a reservoir held at room temperature poses a problem . for the multistage zeeman decelerator used in our experiments , the deceleration - solenoid geometry and the 8@xmath20s switching times of the pulsed deceleration magnetic fields limit the maximal initial velocity of the particles to be decelerated to about 700 m/s @xcite . the common approach to produce low - velocity supersonic beams consisting of seeding the species of interest in a heavy rare gas is not an option in the case of @xmath21 and @xmath14 because these metastable species carry sufficient internal energy to cause the penning ionization of all other rare gases @xcite . the operation of the pulsed - valve assembly , including discharge electrodes , at cryogenic temperatures is therefore the only viable option to reduce the initial velocity of the supersonic beam to below 700 m/s . to produce slow and velocity - tunable beams of metastable helium molecules at low speed , we have combined a pulsed cryogenic supersonic beam source with a multistage zeeman decelerator in the experimental setup depicted schematically in fig . [ fig : setup ] ( a ) . the setup consists of the pulsed - valve assembly with discharge electrodes , the actual multistage zeeman decelerator , and a detection region where the metastable molecules are photoionized with uv laser radiation and the ions or electrons are extracted in the direction perpendicular to the beam propagation axis toward a microchannel - plate ( mcp ) detector . the zeeman decelerator was described in detail in ref . consequently , only the beam source is presented in detail in the following description of the experimental setup and procedure . rotational ground state of @xmath22 @xmath23 . the low - field - seeking states , for which the deceleration pulse sequence is optimized , are indicated in red . ] the pulsed supersonic beam is created by expanding helium gas into vacuum from a reservoir held at high stagnation pressure ( typically 26bar ) through the 250@xmath20m - diameter ceramic orifice of a modified even - lavie valve @xcite . to efficiently cool the valve , its entire housing , originally made of stainless steel , was replaced by copper components . the valve assembly is mounted onto the copper baseplate of a @xmath150.7l cylindrical container which can be filled with cryogenic liquids , either liquid nitrogen for operation at 77k , or liquid helium for operation down to 4k . the temperature of the valve assembly is monitored with a silicon diode ( lakeshore dt-670 ) directly attached to it . throughout this paper , we refer to the temperature measured with this sensor as the `` source temperature '' . starting with the valve assembly at room temperature , it takes about two hours until the valve and discharge operate stably at about 10k . once the desired final temperature is reached , the nozzle - assembly temperature is actively stabilized with a resistive heater attached to the valve . with feedback from a closed - loop proportional - integral - derivative ( pid ) controller ( lakeshore 336 ) , the source temperature fluctuates by less than 0.1k and conditions remain stable for several hours . the metastable helium molecules are generated by striking a discharge and forming a plasma in the expansion region . for this purpose , the expanding gas passes through a 10mm - long , 1.6mm - diameter channel defined by two cylindrical electrodes directly attached to the front plate of the valve and separated by an electrically insulating sapphire spacer ( 2 mm length , inner diameter of cylindrical channel 2 mm ) . the discharge is ignited by applying a typically 5@xmath20s long , @xmath24v pulse to the front electrode as soon as the density of the helium beam in the channel is sufficiently high . in the discharge , the metastable @xmath25 @xmath26 and @xmath27 states of atomic helium are populated , and the metastable helium dimer molecules are formed . to facilitate ignition of the discharge during the short duration of the gas pulse ( typically a few tens of microseconds ) , the discharge is seeded with electrons emitted from a hot tungsten filament @xcite and attracted toward the channel by a @xmath28v bias on the front electrode . without the filament , ignition of the discharge required much higher electric tensions , which unnecessarily heated the beam . the discharge and nozzle operation parameters ( electric tension , timing , stagnation pressure ) were optimized by monitoring the yield of @xmath14 with the methods described in section [ sec : setup : detection ] . we have also tested other discharge - electrode designs ( e.g. , high electric tension applied to a sharp tip in the expansion region @xcite , cylindrical electrodes surrounding a glass capillary ) and found the best yield of @xmath14 with the channel described above . all discharge configurations produced beams of metastable helium atoms , but @xmath29 was only observed in the configuration described above . in the experiment , we observed a strong influence of the discharge electrode temperature on the final velocity of the beam of metastable helium atoms and molecules . this observation may be explained by the discharge electrodes being part of the channel through which the gas expands , thus constituting a virtual gas reservoir for the expansion and beam - formation process . to cool this critical part of the setup to the lowest possible temperatures , the discharge assembly was mechanically connected and thermally linked to the copper front plate of the valve body . moreover , the valve and discharge assembly were surrounded by a thermal shield cooled with liquid nitrogen , which served the purpose of minimizing the heat load of thermal radiation from the hot filament and from components of the vacuum chamber at room temperature . for fast data acquisition in spectroscopic experiments , the source can be operated at the full repetition rate of 25hz of the nd : yag - pumped dye - laser system used to detect the metastable molecules . in this case , a 650l / s ( nominal pumping speed for he ) turbo molecular pump sufficed to maintain the pressure in the source vacuum chamber in the @xmath30 mbar range . after the formation of he@xmath11 molecules , the supersonic beam passed through the 2mm - diameter hole of a skimmer , which separates the source chamber from the multistage zeeman decelerator or the photoelectron spectrometer used to characterize the beam properties and guarantees efficient differential pumping . the multistage zeeman decelerator used in the present work consists of an array of deceleration solenoids to which current pulses are applied and was described in ref . the deceleration results from the force @xmath31 that acts on a paramagnetic atom or molecule with magnetic moment @xmath32 and zeeman energy shift @xmath33 in the presence of an inhomogeneous magnetic field @xmath34 . particle - trajectory simulations indicated that 55 deceleration stages operated at a current of 300a suffice to fully decelerate a beam of @xmath14 with an initial velocity of about 500 m/s . the 55 solenoids ( length 7.2 mm , 64 windings in four layers , 7.0 mm inner diameter ) used as deceleration stages were grouped in two modules of each twelve stages and one module of 31 stages . these modules were separated by pumping towers , as depicted in fig . 2 of ref . @xcite . the zeeman effect in the lowest rotational level @xmath35 of the metastable @xmath10 state of he@xmath9 is depicted in fig . [ fig : setup ] ( b ) , where the low - field - seeking magnetic sublevels ( @xmath36 , drawn in red ) suitable for deceleration experiments have a magnetic moment corresponding to two bohr magnetons . because the spin - rotation coupling constant of low vibrational levels of the @xmath10 state of he@xmath9 is very small @xcite , the zeeman effect in higher rotational levels follows the same pattern as in the @xmath37 level ( see also discussion in section [ sec : conclusion ] ) . to generate beams of he@xmath11 with velocities in the range 100150 m/s , the decelerator was operated at a repetition rate of 8.3hz by applying 250a current pulses to the deceleration solenoids , resulting in maximal magnetic fields on the decelerator axis of about 1.8 t and zeeman shifts of about 1.6@xmath38 for the low - field - seeking magnetic sublevels . the pulse sequences were precalculated to reach the desired final velocities for molecules in these sublevels . optimal conditions were obtained for phase angles between 35@xmath39 and 45@xmath39 , as in our previous deceleration experiments with deuterium atoms @xcite , metastable ne @xcite , and o@xmath9 @xcite . the electric discharge described in section [ sec : setup : cryo ] leads to the production of metastable he atoms in the @xmath25 @xmath26 and @xmath27 states and of metastable @xmath0 molecules in the @xmath10 state . because both atoms and molecules are entrained at the same velocity in the supersonic expansion , they can not be distinguished on the basis of their times of flight to a detector placed along the propagation axis of the supersonic beam . to observe the atoms and the molecules separately , they were photoionized with uv laser radiation produced by second - harmonic generation of a nd : yag - pumped pulsed dye laser and the photoions were extracted through a field - free time - of - flight tube toward a microchannel - plate detector using an electric - field pulse in the range 10002000v / cm . the uv wave number ( 34467@xmath40 ) was chosen to be above the ionization threshold of the metastable @xmath10 state of he@xmath9 , in a region free of autoionization resonances , and also above that of the metastable ( 1s)(2s ) @xmath41 state of he . consequently , both he@xmath42 and he@xmath5 were observed in the ion time - of - flight ( tof ) spectrum , as is illustrated in fig . [ fig : iontof ] . the enhanced electronic noise visible in the time - of - flight spectrum between 1.01.25@xmath20s corresponds to the time at which the extraction field was applied , i.e. , 1@xmath20s after the laser pulse . and extracting the @xmath43 and @xmath44 ions with a pulsed extraction field applied 1@xmath20s after the laser pulse . the electronic noise between 1.01.25@xmath20s is an artifact originating from the electric - field switch - on process . ] to measure the photoionization spectra of he@xmath11 , the signal was integrated over a time window centered at the arrival time of @xmath44 and recorded as a function of the excitation wave number , which was calibrated by recording the optogalvanic spectrum of ne and the transmission through an etalon simultaneously with each spectrum . high - resolution photoelectron spectra of @xmath29 were recorded using the technique of pfi - zeke photoelectron spectroscopy @xcite following the same procedure as in our earlier study of the he@xmath45 photoionizing transition @xcite . the internal state distribution of the metastable molecules was deduced from the rotational structure of the photoelectron spectra and the analysis of the rydberg series observed in the photoionization spectra , as explained in section [ sec : results : source ] . the flight times of the metastable molecules from the point where they were generated in the discharge to the point where they were photoionized by the uv laser ( called `` neutral - tof spectra '' hereafter ) were obtained by monitoring the resulting @xmath44 signal as a function of the delay time between the ignition of the discharge and the photoionization laser pulse . from the measured tof distributions , the velocity distribution of the metastable molecule in the beam can be determined . the @xmath29 neutral - tof spectra obtained for nozzle - assembly temperatures of 300k , 77k and 10k by varying the delay between the discharge ignition pulse and the firing of the ionization laser are compared in fig . [ fig : metastabletof ] . analyzing these neutral - tof distributions following the procedure described in ref . @xcite enables one to extract , in each case , the average velocity of the molecular beam and the corresponding standard deviation , from which the translational temperature of the he@xmath11 sample can be estimated . from fits to the experimental time - of - flight profiles , the mean velocity and velocity spread of the beam are determined to be @xmath46=(1875 , 975 , 530)m / s and @xmath47=(95 , 65 , 60)m / s for source temperatures of ( 300 , 77 , 10)k . when the valve body is stabilized to 10k , the beam properties are sensitive to the discharge conditions , which results in day - to - day variations of the mean velocity by @xmath48 m/s . the results of this analysis are summarized in table [ tab : velocity ] and indicate that the main effect of cooling the nozzle is a strong reduction of the average velocity of the he@xmath11 molecules , the standard deviation being almost independent of the temperature of the nozzle assembly , particularly below 100k . neutral - tof spectra obtained for temperatures of the nozzle assembly of 300k , 77k , and 10k . the red curves correspond to fits to the experimental data that led to the parameters summarized in table [ tab : velocity ] . see text for details . ] .parameters describing the velocity distribution of the beam of @xmath29 obtained from an analysis of the neutral - tof spectra shown in fig . [ fig : metastabletof ] . [ cols= " < , > , > , > " , ] [ tab : velocity ] the internal - state distribution of the molecular sample in the supersonic beam was determined from the rotational structure of the photoelectron spectrum of the origin band of the @xmath49 photoionizing transition . the photoelectron spectra obtained for nozzle - assembly temperatures of 10k and 300k are displayed in fig . [ fig : zeke ] ( a ) and ( b ) , respectively . they were recorded using the technique of pfi - zeke photoelectron spectroscopy @xcite by monitoring the delayed pulsed - field ionization of high rydberg states ( principal quantum number @xmath50 ) located below the successive ionization thresholds of @xmath29 as a function of the uv laser wave number . as explained in ref . @xcite , removal of the @xmath51 electron leads to a dominant q - type rotational branch in the pfi - zeke photoelectron spectrum of @xmath29 , corresponding to transitions between rotational levels of the neutral molecule and the ion having the same rotational quantum number , i.e. , @xmath52 ( @xmath8 and @xmath53 are the rotational quantum numbers of @xmath29 and @xmath44 , respectively ) . the positions of the q - type transitions of the @xmath54 band for @xmath8 values between 1 and 21 are indicated along the upper assignment bar at the top of fig . [ fig : zeke ] ( a ) . weak s - type lines ( i.e. , lines corresponding to @xmath55 ) are also observable in fig . [ fig : zeke ] , and so are two strong and sharp autoionization resonances which strongly enhance the intensity of the o(3)- and o(5)-branch lines as a result of rotational channel interactions , as discussed in detail in ref . the very weak lines corresponding to the lower assignment bar form the q - type branch of the @xmath56 band . the observation of these lines indicates that the @xmath57 level of the a @xmath58 state of he@xmath9 is weakly populated in the supersonic beam . the higher signal - to - noise ratio obtained at a nozzle - assembly temperature of 10k indicates that the density of he@xmath11 is almost an order of magnitude larger than when the nozzle assembly is operated at room temperature . whereas the rotational - line - intensity distribution of the spectrum recorded following expansion from the 300k nozzle is well described by a rotational temperature of about 150k , the spectrum of the @xmath29 sample obtained with the 10k nozzle assembly is characterized by a bimodal rotational intensity distribution , with a cold component corresponding to a temperature of about 150k for levels with @xmath59 and a much hotter ( nonthermal ) component corresponding to rotational levels in the range @xmath60 with a maximum at @xmath61 . the 10k nozzle assembly thus leads to a rotationally warmer sample than the 300k nozzle assembly . this somewhat counter - intuitive observation may be explained by the large rotational constant ( @xmath62 , see ref . @xcite ) of @xmath29 and the fact that only odd-@xmath8 rotational levels of the @xmath10 state are allowed by the generalized pauli principle : the spacing between successive rotational levels rapidly becomes larger than the thermal collision energy when the nozzle assembly is held at low temperature and , consequently , cooling of the rotational degrees of freedom becomes very inefficient . the broad range of rotational states that are populated in the slow beam produced with the cold nozzle opens up the possibility to study levels with rotational energy exceeding 3000@xmath40 by high - resolution spectroscopy under the collision - free conditions of a supersonic beam . detected 60 mm beyond the last stage of the decelerator . the photoionization laser was set to 34467@xmath38 , i.e. , above all relevant ionization thresholds , to avoid the artificial enhancement of the photoionization signal of molecules in specific rotational levels of the @xmath10 state . the black trace with a broad tof distribution centered at 500 m/s corresponds to the undecelerated beam . the three distributions presented at times of flight between 2.9ms and 3.3ms were obtained using deceleration pulse sequences designed to decelerate molecules with initial velocities of 505 m/s , 500 m/s , and 495 m/s , respectively , at a phase angle of 35@xmath39 . ] when the nozzle assembly and the discharge electrodes are cooled down to about 10k , the he@xmath11 molecules are entrained in a supersonic beam with a mean velocity of @xmath63 m/s , which is well below the maximal initial velocity tolerated by our zeeman decelerator . as mentioned in section [ sec : setup : zeeman ] , 55 deceleration stages grouped in three modules suffice to decelerate he@xmath11 to final velocities of 100 m/s . time - of - flight profiles of @xmath29 detected at the exit of the zeeman decelerator , i.e. , at a distance of about 1 m from the nozzle orifice , are displayed in fig . [ fig : decel ] . these profiles were obtained by the procedure ( uv - laser photoionization mass spectrometry ) used to characterize the beam source and discussed in section [ sec : results : source ] . to prevent the artificial enhancement of the photoionization signal of molecules in specific rotational levels of the @xmath10 state , the uv - laser wave number was kept fixed at 34467@xmath38 , in a structureless region of the photoionization spectrum . in fig . [ fig : decel ] , the black trace , with a broad peak of @xmath29 molecules centered at a delay of 2.05ms , represents the neutral tof distribution recorded without switching any of the deceleration solenoids and corresponds to @xmath29 molecules propagating at the original beam velocity of @xmath15500 m/s through the entire decelerator . when running the decelerator with current pulses of 250a and a phase angle of 35@xmath39 , the tof profiles observed in the region between 2.9ms and 3.3ms were obtained . the profiles with maxima at 2.95ms ( red trace ) , 3.1ms ( black trace ) and 3.25ms ( blue trace ) were obtained under identical conditions except that the pulse sequences used for the deceleration were precalculated to decelerate metastable molecules with initial velocities of 505 m/s , 500 m/s , and 495 m/s , respectively . analysis of the time - of - flight profiles with the methods described in ref . @xcite indicates that the final velocities are 150 m/s , 135 m/s , and 120 m/s , respectively , in agreement with the values predicted by numerical particle - trajectory simulations . the structure of the peaks corresponding to the decelerated molecules is a consequence of the evolution of the phase - space distribution in the decelerator @xcite . the neutral - tof distributions presented in fig . [ fig : decel ] demonstrate the possibility to tune the final velocity of the decelerated beam in the range between 100 and 150 m/s . although the present setup allows for the deceleration to lower final velocities in a straightforward way , either by operating the decelerator at a larger phase angle or by applying larger deceleration magnetic fields , the detection of these slow molecules becomes increasingly difficult . in particular , the rather large distance of 60 mm between the last solenoid of the decelerator and the detection volume inevitably reduces the particle density by transverse defocussing and longitudinal dispersion effects , which play a larger role at lower final velocities @xcite . for the efficient generation of beams with velocities below 100 m/s , deceleration pulse sequences with optimized phase angles , such as those already used for trap - loading schemes @xcite , might be useful to focus and bunch the molecules into the detection region . the maximum intensity of the peak in fig . [ fig : decel ] corresponding to the decelerated he@xmath11 molecules is only a factor of three weaker than the maximum intensity of the undecelerated beam . consequently , the internal - state distribution of the decelerated sample can be characterized by spectroscopic methods with almost the same sensitivity as achieved for the undecelerated beam ( see section [ sec : results : source ] ) . unfortunately , it was not possible to record pfi - zeke photoelectron spectra in the region beyond the last coil of the decelerator . indeed , the detection of the photoelectrons following extraction in the direction parallel to the beam propagation axis was hindered by the background signal originating from metastable atoms and molecules and electrons generated by the discharge . the stray magnetic fields along the decelerator axis further prevented the detection of electrons extracted in the direction perpendicular to the beam propagation axis . the internal - state distribution of the decelerated sample had therefore to be determined from the photoionization spectrum of @xmath29 , which is dominated by contributions from autoionization resonances in the vicinity of the ionization thresholds ( see fig . 9 of reference @xcite ) . the photoionization spectra recorded in the region 3412534185@xmath64 for decelerated and undecelerated @xmath29 beams are compared in fig . [ fig : spectrum ] . these spectra are dominated by @xmath65 rydberg series @xcite . whereas the @xmath66 series , for which the character of the rydberg electron is @xmath67 , are regular and well described by rydberg s formula , the @xmath68 and @xmath69 series are of mixed @xmath70 and @xmath71 character and are strongly perturbed by rotational channel interactions with the @xmath72 and @xmath73 series , respectively . for clarity , the assignments are only given for the regular series in fig . [ fig : spectrum ] and the lines belonging to the perturbed series , which have also been assigned using the multi - channel - quantum - defect - theory models described in refs . @xcite , are marked by asterisks . the unambiguous assignment of transitions originating from rotational levels of metastable @xmath29 with @xmath8 values between 1 and 9 demonstrates that many rotational levels are populated in the decelerated sample . moreover , the almost identical appearance of the spectra of the decelerated and undecelerated samples presented in fig . [ fig : spectrum ] implies that the multistage zeeman deceleration process is not rotationally state selective in he@xmath11 . this observation stands in stark contrast to what is observed in the multistage zeeman deceleration of @xmath74 , for which the decelerated sample was found to consist exclusively of molecules in the @xmath75 magnetic sublevel of the @xmath76 fine - structure component of the @xmath77 rotational ground state @xcite . the reasons for this very different behavior lie in the very different relative magnitudes of the rotational constants and spin - rotation splittings of these two molecules , and in the fact that the paschen - back limit of the zeeman effect is reached at much lower magnetic fields in the metastable @xmath10 state of @xmath0 than in the @xmath18 ground state of @xmath74 , as is discussed in more detail in the next section . ground state of @xmath74 ( left column ) and in the @xmath10 state of @xmath0 ( right column ) . top row : zeeman energy of the rotational states @xmath78 in the magnetic - field range between 0 and 20 t . bottom row : zeeman energy in the magnetic - field range where the zeeman shifts are comparable to the spin - rotation splittings . note the different scales used in these panels . ] in this article , the design and operational characteristics of a source of slow , velocity - tunable beams of metastable @xmath0 molecules ( @xmath29 ) have been presented . the molecules are formed in a discharge , entrained in a supersonic expansion of he , and decelerated to low velocity in a multistage zeeman decelerator . at the end of the decelerator , the density of @xmath29 is sufficiently high for measurements of the rydberg spectrum of @xmath0 by photoionization spectroscopy . @xmath0 in its metastable @xmath10 state has almost 20ev of internal energy , which is enough to ionize any atom or molecule except he . consequently , it is not possible to form a supersonic beam of @xmath29 seeded in any carrier gas other than atomic helium . a supersonic beam of pure he expanding from a room - temperature reservoir held at high stagnation pressure , however , has a mean velocity of @xmath152000 m/s , which is beyond the range of velocities for which deceleration is possible with our multistage zeeman decelerator @xcite . indeed , such a beam propagates by more than the 7.2 mm length of the solenoids used for deceleration during the switch - off time of about 8@xmath20s of the current pulses . to overcome this limitation , we have developed a source which can be cooled to 10k and at the exit of which @xmath29 is generated in a discharge . @xmath29 is then entrained by the he carrier gas in a supersonic beam having a mean velocity of about 500 m/s . phase - stable deceleration to velocities as low as 100 m/s could be achieved using a zeeman decelerator consisting of 55 stages . the velocity distribution of the decelerated molecules was measured by time - of - flight spectrometric methods and their internal - state distribution was determined from the analysis of well - resolved rydberg series converging on the rovibrational levels of the @xmath12 ground electronic state of @xmath44 . the low translational temperature of the decelerated sample ( less than 100mk ) and the broad range of populated rotational levels ( @xmath8 ranging from 1 to 21 ) makes this source ideally suited to measurements of the rydberg spectrum of @xmath0 , of the ionization energy of @xmath29 and of the rovibrational structure of @xmath44 by high - resolution photoionization and photoelectron spectroscopic methods . unexpectedly , the population of metastable molecules in the supersonic expansion were found to be distributed over many more rotational levels when the valve was cooled down to low temperature than when it was operated at room temperature . we interpret this observation as arising from the fact that the rotational spacings in @xmath29 are much larger than the thermal collision energy when the nozzle assembly is held at low temperature , which effectively hinders rotational cooling . the broad range of rotational levels in the supersonic expansion enabled us to study how the zeeman deceleration process affects the rotational - state distribution . in contrast to o@xmath9 , for which multistage zeeman deceleration led to the formation of slow beams of molecules in a single magnetic sublevel of a single spin - rotational level of the ground electronic state @xcite , we did not observe any rotational state selectivity in the multistage zeeman deceleration of @xmath29 . the reason for the very different behaviors observed in o@xmath9 and @xmath29 lies in the different nature of the zeeman effect in these two molecules , which itself results from the fact that @xmath29 has a much larger rotational constant than o@xmath9 ( @xmath62 @xcite vs. @xmath79 @xcite ) and spin - rotational splittings more than two orders of magnitude smaller than @xmath74 @xcite . the zeeman effect in the ground vibrational level of @xmath74 @xmath18 and @xmath0 @xmath10 are compared in fig . [ fig : zeemancomparison ] . in @xmath74 , the paschen - back regime is not reached at the magnetic fields up to 2 t used in our experiments because of the large fine - structure splitting of the rotational levels . moreover , interactions between magnetic sublevels of the same @xmath80 values but belonging to different rotational levels cause strong nonlinearities and avoided crossings already below 5 t . in @xmath29 , the paschen - back regime is already reached at fields of less than 0.1 t , and the large rotational spacings prevent any avoided crossings even at fields in excess of 15 t . consequently the low - field - seeking magnetic sublevels of the different rotational levels of @xmath29 are subject to almost identical zeeman shifts at the fields used in our deceleration experiments , which makes the deceleration process completely insensitive to the degree of rotational excitation . consideration of the zeeman effect in the rotational levels of the @xmath10 of @xmath0 , however , reveals that none of the magnetic sublevels of the spin - rotational levels with @xmath6 are low - field seeking ( see fig . [ fig : zeemancomparison ] ( d ) ) . multistage zeeman deceleration of @xmath29 therefore acts as a filter which selects the two spin - rotational components with @xmath81 and rejects the third . this property will represent an advantage in studies of the very congested rydberg spectrum of @xmath0 . we thank richard maceiczyk and patrick seewald for their support in the early phase of this work . this work is supported by the swiss national science foundation ( project 200020 - 149216 ) and the european research council advanced grant program ( project 228286 ) . m. m. thanks eth zrich for the support through an eth fellowship .	@xmath0 molecules in the metastable @xmath1 state have been generated by striking a discharge in a supersonic expansion of helium gas from a pulsed valve . when operating the pulsed valve at room temperature , 77k , and 10k , the mean velocity of the supersonic beam was measured to be 1900 m/s , 980 m/s , and 530 m/s , with longitudinal velocity distributions corresponding to temperatures of 4k , 1.9k , and 1.8k , respectively . the characterization of the population distribution among the different rotational levels of the @xmath1 state by high - resolution photoelectron and photoionization spectroscopy indicated a rotational temperature of about 150k for the beam formed by expansion from the room - temperature valve and a bimodal distribution for the beam produced with the valve held at 10k , with rotational levels up to @xmath2 being populated . a 55-stage zeeman decelerator operated in a phase - stable manner in the longitudinal and transverse dimensions was used to further reduce the beam velocity and tune it in the range between 100 and 150 m / s . the internal - state distribution of the decelerated sample was determined by recording and analyzing the photoionization spectrum in the region of the lowest ionization threshold , where it is dominated by resonances corresponding to autoionizing @xmath3p rydberg states belonging to series converging to the different rotational levels of the @xmath4 ground state of he@xmath5 . the deceleration process did not reveal any rotational state selectivity , but eliminated molecules in spin - rotational sublevels with @xmath6 from the beam , @xmath7 and @xmath8 being the total and the rotational angular momentum quantum number , respectively . the lack of rotational state selectivity is attributed to the fact that the paschen - back regime of the zeeman effect in the rotational levels of the @xmath1 state of @xmath0 is already reached at fields of only 0.1 t .
inspired by ultracold atomic physics , nuclear physics and color superconductivity , the fermion pairing between different species with mismatched fermi surfaces prompted great interest@xcite in recent experimental and theoretical studies . in conventional fermion superfluid the ground state is well described by the bcs theory , while for asymmetric fermion superfluid with double - species pairing the phase structure is much more rich and the pairing mechanism is not yet very clear . various exotic phases have been suggested in the literatures , such as the sarma phase@xcite , the fulde - ferrel - larkin - ovchinnikov(fflo ) phase@xcite , the phase with deformed fermi surfaces@xcite , and the phase separation@xcite in coordinate space . for asymmetric fermion superfluid , one of the most surprising phenomena is the magnetic instability , @xmath0 , when the asymmetry characterized by the chemical potential mismatch is larger than the condensate of cooper pair , the bcs superfluid is unstable under the perturbation of an external magnetic field . this instability leads to the ground state to be a fflo phase . for color superconductivity , a similar but more complicated phenomenon is also found , namely the chromomagnetic instability@xcite . one of the mechanisms to cure the magnetic instability is to introduce another pairing channel . in this paper , we propose a simple model including both double- and single - species pairings and investigate how the single - species pairing influences the fflo state . our model can be used to describe a general fermion superfluid system where fermions from the same species can form cooper pairs . in ultracold atom gas like @xmath1li and @xmath2k , there exist pairings between different elements and between different states of the same element@xcite . in color superconductivity phase of dense quark matter@xcite , the quarks of different flavors can form total spin zero cooper pairs and the quarks of the same flavor can combine into total spin one pairs too . in neutron stars@xcite , proton - proton , neutron - neutron and neutron - proton pairs are all possible . recently discovered high temperature superconductivity of mgb@xmath3@xcite can also be well described by an extended two - band bcs theory where the electrons from the same energy bands form cooper pairs . we consider a fermion system containing two species @xmath4 and @xmath5 with masses @xmath6 and @xmath7 and chemical potentials @xmath8 and @xmath9 satisfying @xmath10 . the system can be described by the hamiltonian @xmath11 \nonumber\\ & & -\sum_{{\bf p},{\bf k}}\left[\frac{g}{2}\sum_{\sigma\sigma'}\hat a_{\bf p}^{\sigma+}\hat b_{-{\bf p}}^{-\sigma+}\hat b_{-{\bf k}}^{-\sigma'}\hat a_{\bf k}^{\sigma'}+g_a\hat a_{\bf p}^{\uparrow+}\hat a_{-{\bf p}}^{\downarrow+}\hat a_{-{\bf k}}^\downarrow\hat a_{\bf k}^\uparrow + g_b\hat b_{\bf p}^{\uparrow+}\hat b_{-{\bf p}}^{\downarrow+}\hat b_{-{\bf k}}^\downarrow\hat b_{\bf k}^\uparrow\right],\end{aligned}\ ] ] where @xmath12 and @xmath13 are fermion momenta , @xmath14 and @xmath15 are the annihilation and creation operators , the coupling constants @xmath16 and @xmath17 are positive to keep the interactions attractive , @xmath18 and @xmath19 are the non - relativistic dispersion relations , and in continuous limit we simply replace @xmath20 with @xmath21 . the ( pseudo-)spin @xmath22 has been introduced to keep our model satisfying the pauli rule explicitly . other inner structures like isospin , flavor and color are neglected since they are not essential for our purpose . the hamiltonian has the symmetry @xmath23 with the element @xmath24 defined as @xmath25 and @xmath26 . in the framework of mean field approximation , after taking a bogoliubov - valatin transformation from fermions @xmath4 and @xmath5 to quasi - particles @xmath27 and @xmath28 , the thermodynamic potential can be expressed in terms of the quasi - particles , @xmath29,\end{aligned}\ ] ] where @xmath30 are the corresponding @xmath4-@xmath4 , @xmath5-@xmath5 and @xmath4-@xmath5 pairing condensates , and @xmath31 are the quasi - particle energies with @xmath32/2,\nonumber\\ \left(\epsilon_\delta^2\right)^2&=&\delta^2\left[\left(\epsilon_a-\epsilon_b\right)^2+\left(\delta_a-\delta_b\right)^2\right].\end{aligned}\ ] ] @xmath33 and @xmath34 correspond , respectively , to the spontaneous symmetry breaking patterns @xmath35 and @xmath36 , and @xmath37 means the breaking pattern @xmath38 with the element @xmath39 defined as @xmath40 and @xmath41 . for simplicity , we have considered all the condensates as real numbers . to study the fflo state , we modify the definitions of the fermion dispersions as @xmath42 with @xmath43 corresponding to spin - up and @xmath44 to spin - down , where @xmath45 are the fflo momenta of condensates of @xmath46 pairing , which together with the condensates are self - consistently determined by the coupled set of gap equations @xmath47 and the minimum thermodynamic potential . note that the fflo state we considered here is the simplest form where the cooper pairs in coordinate space have the plane wave forms @xmath48 we will choose @xmath49 , such a choice can avoid possible dynamic instability due to the different superflow velocities of @xmath50 and @xmath51 channels@xcite . obviously , the translational symmetry and rotational symmetry in the fflo state are spontaneously broken . to have a simple analytical expression for the spectral of quasi - particles , we set @xmath52 . in this case , the thermodynamic potential reads , @xmath53,\end{aligned}\ ] ] where @xmath54 are the quasi - particle energies with the notations @xmath55/2,\nonumber\\ \epsilon_a^2 & = & \left[\left(\epsilon_a^+\epsilon_a^-+\delta_a^2+\delta^2\right)-\left(\epsilon_b^+\epsilon_b^-+\delta_b^2+\delta^2\right)\right]/2,\nonumber\\ \delta\epsilon&=&(\epsilon_a^+-\epsilon_a^-)/2=(\epsilon_b^+-\epsilon_b^-)/2,\nonumber\\ \left(\epsilon_\delta^2\right)^2&=&\delta^2\left[(\epsilon_a^+-\epsilon_b^+)(\epsilon_a^--\epsilon_b^-)+(\delta_b-\delta_a)^2\right].\end{aligned}\ ] ] the magnetic stability , namely the stability of the homogeneous superfluid against the pair momentum fluctuations is characterized by the superfluid density @xmath56 positive @xmath57 means stable homogeneous superfluid and negative @xmath57 indicates possible fflo state . in this sense @xmath58 plays the role of an external magnetic potential @xmath59 . in the end of this section , we calculate the gapless nodes for the case with zero fflo momentum where at least one of @xmath60 or @xmath61 crosses the momentum axis , @xmath62 there are two classes of gapless solutions . 1 ) only @xmath63 is nonzero . the gapless mode happens at momenta @xmath64 where we have introduced the average chemical potential @xmath65 and the chemical potential mismatch @xmath66 . it is clear that the condition to have the gapless mode is @xmath67 . 2 ) all the three condensates are zero or only one of @xmath68 and @xmath69 is nonzero . in these cases , we have @xmath70 or @xmath71 or @xmath72 which mean real fermion excitations exactly at the fermi surfaces . since the model is non - renormalizable , we introduce in the numerical calculations a cutoff @xmath73 with @xmath74 and the average fermi momentum @xmath75 . we choose @xmath76 with @xmath77 and @xmath78 being the @xmath79-wave scattering lengths . we have checked that in the parameter region of @xmath80 and @xmath81 there is no qualitative change in the phase diagrams . plane at fixed @xmath82 with @xmath83 being the symmetric gap at @xmath84 . the upper panel is with only double - species pairing , and in the bottom panel the single - species pairing is included as well . , title="fig:",width=302 ] plane at fixed @xmath82 with @xmath83 being the symmetric gap at @xmath84 . the upper panel is with only double - species pairing , and in the bottom panel the single - species pairing is included as well . , title="fig:",width=302 ] the phase diagram in @xmath85 plane is shown in fig.[fig1 ] . the upper panel is in the familiar case without single - species pairing@xcite . the homogeneous bcs state can exist at low temperature and low mismatch , and the inhomogeneous fflo state survives only in a narrow mismatch window . when the single - species pairing is included as well , see the bottom panel of fig.[fig1 ] , the inhomogeneous fflo state of @xmath4-@xmath5 pairing is eaten up by the homogeneous superfluid of @xmath4-@xmath4 and @xmath5-@xmath5 pairings at low temperature , just as we expected , and survives only in a small triangle at high temperature . the phase transition from the @xmath4-@xmath5 pairing superfluid to the @xmath4-@xmath4 and @xmath5-@xmath5 pairing superfluid is of first order . note that for systems with fixed chemical potentials there is no mixed phase of double- and single - species pairings , and the situation is similar to a three - component fermion system@xcite . we have investigated the phase structure of an asymmetric two - species fermion superfluid with both double- and single - species pairings . since the attractive interaction for the single - species pairing is relatively weaker , it changes the conventional phase diagram at low temperature . in any system with chemical potential imbalance , the double - species pairing in fflo state is replaced by the single - species pairing . in the region with single - species pairing , the interesting gapless superfluid is washed out and all fermion excitations are fully gapped . we should note that in this paper we considered only the grand canonical ensemble . for the case of canonical ensemble , the phase diagram becomes much more rich@xcite . + + * acknowledgments : * the work was supported by the grants nsfc10575058 , 10425810 and 10435080 . 20 m.w.zwierlein @xmath86 , science * 311 * , 492(2006 ) . g.b.partidge @xmath86 , science * 311 * , 503(2006 ) . y.shin @xmath86 , phys . lett . * 97 * , 030401(2006 ) . d.t.son and m.a.stephanov , phys . rev . * a74 * , 013614(2006 ) . j.carlson and s.reddy , phys . rev . lett . * 95 * , 060401(2005 ) . l.he , m.jin and p.zhuang , phys . rev . * b73 * , 214527(2006 ) . m.iskin and c.a.r.sa de melo , phys * b74 * , 144517(2006 ) ; phys . * 97 * , 100404(2006 ) . a.bulgac , m.m.forbes and a.schwenk , phys . lett . * 97 * , 020402(2006 ) . t.ho and h.zhai , cond - mat/0602568 . i.giannakis , d.hou , m.huang and h.ren , cond - mat/0603263 . h.hu and x.liu , phys a73 * , 051603(r)(2006 ) . k.machida , t.mizushima and m.ichioka , phys . 97 * , 120407(2006 ) . h.caldas , cond - mat/0605005 . g.sarma , j.phys.chem.solids * 24 * , 1029(1963 ) . i.shovkovy and m.huang , phys b564 * , 205(2003);m.huang and i.shovkovy , phys.rev.d 70 , 094030(2004 ) . w.liu and f.wilczek , phys . lett . * 90 * , 047002(2003 ) . p.fulde and r.ferrel , phys . rev . * a135,550(1964 ) ; a.larkin and y.ovchinnikov , sov . phys . jetp 20,762(1965 ) . h.muther and a.sedrakian , phys . lett . 88 * , 252503(2002 ) . a.sedrakian @xmath86 , phys . rev . * a72 * , 013613(2005 ) . p.bedaque , h.caldas and g.rupak , phys . * 91 * , 247002(2003 ) . h.caldas , phys . rev . * a69 * , 063602(2004 ) . x.huang,x.hao and p.zhuang , new j. phys . * 9,375 ( 2007 ) , [ arxiv : cond - mat/0610610 ] . m.huang , p.zhuang and w.chao , phys . rev . d67 * , 065015(2003 ) ; i.shovkovy , found . 35 * , 1309(2005 ) ; m.huang , int.j.mod.phys.e14 , 675(2005 ) ; h.ren , hep - ph/0404074 . u.lombardo @xmath86 , int.j.mod.phys . * e14,513(2005 ) . m.iavarone @xmath86 , phys . 89 , 187002(2001 ) . f.bouquet @xmath86 , phys . lett . * 89 * , 257001(2002 ) . s.tsuda @xmath86 , phys . * 91 * , 127001(2003 ) . j.geerk @xmath86 , phys . lett . * 94 * , 227005(2005 ) . i.m.khalatnikov , jetp lett . * 17 * , 386(1973 ) . v.p.mineev , sov . phys.jetp * 40 * , 132(1974 ) . y.a.nepomnyashchii , theor . math * 20 * , 904(1974 ) ; sov . jetp * 43 * , 559(1976 ) . s.takata and t.izuyama , prog . . phys . * 41 * , 635(1969 ) . t.paananen , j.martikainen and p.torma , phys . rev . * a73 * , 053606(2006 ) .	we investigate the phase diagram in the plane of temperature and chemical potential mismatch for an asymmetric fermion superfluid with double- and single - species pairings . there is no mixing of these two types of pairings at fixed chemical potential , but the introduction of the single species pairing cures the magnetic instability at low temperature .
let @xmath5 be a graph . we say that @xmath5 is _ connected _ if there is a path between any two distinct vertices of @xmath5 . we denote by @xmath6 , the degree of a vertex @xmath7 in @xmath5 . a graph in which every vertex has the same degree is called a _ regular graph_. if all vertices have degree @xmath8 , then the graph is said to be _ @xmath8-regular_. the _ girth _ of @xmath5 , denoted by @xmath9 , is the length of a shortest cycle in @xmath5 ( we say that @xmath10 if @xmath5 contains no cycle ) . a _ null graph _ is a graph with no edges . a _ forest _ is a graph with no cycle . we denote the complete graph , the path and the cycle of order @xmath11 by @xmath12 , @xmath13 and @xmath14 , respectively . @xmath11-cycle _ to denote the cycle of order @xmath11 , where @xmath15 . cartesian product _ of two graphs @xmath5 and @xmath16 is denoted by @xmath17 . the _ hypercube graph _ @xmath18 is the cartesian product of @xmath19 copies of @xmath20 . + let @xmath0 be a group . we denote the identity element of @xmath0 by @xmath21 . the derived subgroup of @xmath0 is denoted by @xmath22 and @xmath23 , where @xmath11 is a positive integer . for any subgroup @xmath2 of @xmath0 , the intersection of all the conjugates of @xmath2 in @xmath0 is denoted by @xmath24 . let @xmath25 . then the subgroup generated by @xmath26 is denoted by @xmath27 . as usual , @xmath28 , @xmath29 and @xmath30 denote the group of integers modulo @xmath11 , the alternating group and the symmetric group of degree @xmath11 , respectively . for a fixed prime @xmath31 , the _ quasicyclic @xmath31-group _ is denoted by @xmath32 . also the projective special linear group of degree @xmath11 over the field @xmath33 is denoted by psl@xmath34 . + there are several graphs associated with groups , for instance non - commuting graph of a group , intersection graph of subgroups of a group , and subgroup graph of a group . ( see @xcite . ) the subgroup graph of a group @xmath0 is defined as the graph of its lattice of subgroups , that is , the graph whose vertices are the subgroups of @xmath0 such that two subgroups @xmath2 and @xmath3 are adjacent if one of @xmath2 or @xmath3 is maximal in the other . in this article , we introduce and investigate the _ prime index graph _ of @xmath0 , denoted by @xmath1 . it is an undirected graph whose vertices are all subgroups of @xmath0 and two distinct comparable vertices @xmath2 and @xmath3 are adjacent if and only if @xmath35 $ ] or @xmath36 $ ] is prime . clearly , the prime index graph of @xmath0 is a subgraph of the subgroup graph of @xmath0 and whenever @xmath0 is a nilpotent group , see @xcite , then these two graphs are coincide . in follows the prime index graphs of @xmath37 and @xmath38 are given . note that @xmath39 , @xmath40 , for @xmath41 and @xmath42 , for @xmath43 . ( 1,3 ) ( 1,4 ) ( 1,5 ) ( 1,6 ) ( 2,3 ) ( 2,4 ) ( 2,5 ) ( 2,6 ) ( 10,9 ) ( 9,8 ) ( 9,7 ) ( 9,6 ) ( 8,5 ) ( 7,5 ) ( 6,5 ) ( 5,1 ) ( 5,2 ) ( 5,3 ) ( 5,4 ) @xmath44 @xmath45 + here we show that for every group @xmath0 , @xmath1 is a bipartite graph and @xmath46 . we prove that for any finite abelian group @xmath0 , @xmath1 is a regular graph if and only if @xmath1 is a hypercube graph . finally , we study the connectivity of @xmath1 and we show that for every finite solvable group @xmath0 , @xmath1 is a connected graph . among other results , we prove that if @xmath1 is a connected graph and @xmath47 is a normal subgroup of @xmath0 , then both graphs @xmath48 and @xmath49 are connected . in this section , we show that the prime index graph of a group @xmath0 is bipartite . to see this , we prove a stronger result . first we define a directed graph @xmath50 . it is a directed graph whose vertex set is the set of all subgroups of @xmath0 and for every two distinct vertices @xmath2 and @xmath3 , there is an arc from @xmath2 to @xmath3 , whenever @xmath51 and @xmath36=r$ ] , for some positive integer @xmath52 . suppose that @xmath52 is the weight of the arc from @xmath2 to @xmath3 . [ ashtab ] let @xmath53 be a cycle of @xmath50 . then the product of weights of all clockwise arcs of @xmath53 is equal to the product of weights of all counter - clockwise arcs of @xmath53 . let @xmath53 be a cycle of @xmath50 of length @xmath11 . we prove the theorem by induction on @xmath11 . clearly , for @xmath54 the assertion holds . now , suppose that @xmath55 and the assertion is true for every integer @xmath56 , @xmath57 . if @xmath53 contains a directed path @xmath58 of length 2 , such as @xmath59 , then we replace @xmath58 with the path @xmath60 . hence by the induction hypothesis the result holds . otherwise , @xmath53 contains a path of the form @xmath61 . we consider two cases : * case 1. if @xmath62 is not a vertex of @xmath53 , then we replace @xmath63 with the path @xmath64 , where @xmath65=r'$ ] and @xmath66=s'$ ] . note that @xmath67 and so @xmath68 . next , we replace @xmath69 with @xmath70 . thus we find a cycle @xmath71 of length @xmath72 and by the induction hypothesis @xmath73 , where @xmath74 is the product of weights of all clockwise arcs of @xmath71 except the weight of @xmath70 and @xmath75 is the product of weights of all counter - clockwise arcs of @xmath71 except the weight of @xmath76 . hence @xmath77 and so @xmath78 . it is clear that @xmath79 is the product of weights of all clockwise arcs of @xmath53 and @xmath80 is the product of weights of all counter - clockwise arcs of @xmath53 . the result holds . case 2. assume that @xmath62 is a vertex of @xmath53 . clearly , @xmath81 or @xmath82 . with no loss of generality , suppose that @xmath81 . by adding the arc @xmath83 , we find two cycles @xmath71 and @xmath84 of lengths less than @xmath11 . let @xmath66=s'$ ] . assume that the arc @xmath85 is a clockwise arc of @xmath71 . so @xmath85 is a counter - clockwise arc of @xmath84 . now , by the induction hypothesis , @xmath86 , where @xmath87 is the product of weights of all clockwise arcs of @xmath71 except the weight of @xmath85 , @xmath88 is the product of weights of all clockwise arcs of @xmath84 , @xmath89 is the product of weights of all counter - clockwise arcs of @xmath71 , and @xmath90 is the product of weights of all counter - clockwise arcs of @xmath84 except the weight of @xmath85 . thus @xmath91 . the proof is complete . now , we are in a position to prove the following corollary . [ bipartite ] let @xmath0 be a group . then @xmath1 is bipartite . we show that every cycle of @xmath1 is an even cycle . if @xmath1 has a cycle @xmath53 , we may assume that @xmath53 is a cycle in @xmath50 . now , by theorem [ ashtab ] , since all weights are primes , the number of clockwise arcs of @xmath53 is equal to the number of counter - clockwise arcs of @xmath53 . hence @xmath53 is an even cycle . this implies that @xmath1 is a bipartite graph . if @xmath0 is a non - trivial group and @xmath92 , then @xmath93 contains a subgroup of prime index and hence @xmath94 . so @xmath1 is not a null graph . let @xmath0 be a group . then @xmath1 is a complete bipartite graph if and only if @xmath0 is a cyclic group of prime order or @xmath95 , for some primes @xmath31 and @xmath96 . clearly , if @xmath97 , then @xmath98 . also if @xmath95 , then @xmath1 is a complete bipartite graph whose one part contains all subgroups of @xmath0 of orders @xmath31 or @xmath96 and the other part contains @xmath99 and @xmath0 . conversely , assume that @xmath1 is complete bipartite . if @xmath99 and @xmath0 are contained in two different parts of @xmath1 , then @xmath97 , where @xmath31 is a prime number . otherwise , there exists a subgroup @xmath2 of @xmath0 adjacent to both @xmath99 and @xmath0 . thus @xmath95 , for some primes @xmath31 and @xmath96 . the following theorem shows that if @xmath1 contains a cycle @xmath53 , then @xmath100 . [ girth ] let @xmath0 be a group . then @xmath46 . first assume that @xmath0 is finite and @xmath101 , where @xmath102 are distinct primes and @xmath103 are positive integers . suppose that @xmath104 is a sylow @xmath105-subgroup of @xmath0 , for @xmath106 . first assume that @xmath104 contains two distinct maximal subgroups @xmath2 and @xmath3 , for some @xmath107 . since @xmath2 and @xmath3 are normal subgroups of @xmath104 , so @xmath108 . this implies that @xmath109 and hence @xmath104 @xmath2 @xmath110 @xmath3 @xmath104 is a 4-cycle in @xmath1 . so by corollary [ bipartite ] , @xmath100 . next , assume that @xmath104 contains a unique maximal subgroup , for @xmath106 . hence all sylow subgroups of @xmath0 are cyclic . now , by ( ? ? ? * theorem 10.26 ) , @xmath0 is a supersolvable group . if @xmath111 , then @xmath0 has a subgroup @xmath3 of order @xmath112 ( @xcite ) . let @xmath113 be a subgroup of @xmath3 of order @xmath105 , for @xmath114 . hence @xmath99 @xmath3 @xmath116 @xmath99 is a 4-cycle in @xmath1 and so by corollary [ bipartite ] , @xmath100 . if @xmath117 , then @xmath118 . thus @xmath119 and @xmath120 . + now , suppose that @xmath0 is infinite and @xmath1 contains a cycle @xmath53 . it is easy to see that @xmath53 should contain a path of the form @xmath121 @xmath2 @xmath47 , where @xmath2 , @xmath121 and @xmath47 are subgroups of @xmath0 and furthermore @xmath121 and @xmath47 are maximal subgroups of @xmath2 . if both @xmath121 and @xmath47 are normal subgroups of @xmath2 , then @xmath122=[mn : n]=[h : n]$ ] and similarly @xmath123=[h : m]$ ] . thus @xmath2 @xmath121 @xmath124 @xmath47 @xmath2 is a 4-cycle in @xmath1 . now , assume that @xmath121 is not a normal subgroup of @xmath2 . then @xmath121 @xmath2 @xmath125 is a path in @xmath1 , for some @xmath25 . therefore , @xmath126 @xmath127 @xmath128 is a path in @xmath129 . clearly , @xmath127 is a finite group which is not a cyclic @xmath31-group . so by the previous paragraph , @xmath130 and hence @xmath100 . by the proof of the previous theorem , we have the following corollary . if @xmath0 is a finite group or an infinite abelian group , then @xmath1 is a forest if and only if @xmath0 is isomorphic to either @xmath131 or @xmath32 , where @xmath31 is a prime and @xmath11 is a positive integer . suppose that @xmath1 is a forest . if @xmath0 is finite , then by the proof of theorem [ girth ] , @xmath132 , for some prime number @xmath31 and positive integer @xmath11 . if @xmath0 is an infinite abelian group , then @xmath0 is a torsion @xmath31-group . ( note that @xmath133 and if @xmath0 has two elements of orders @xmath31 and @xmath96 , then @xmath134 is a subgroup of @xmath0 , where @xmath135 are distinct primes . ) also by the proof of theorem [ girth ] , every finite subgroup of @xmath0 is cyclic . thus @xmath0 has no non - trivial direct summand . now , by @xcite , @xmath136 , for some prime @xmath31 . clearly , @xmath137 is a disjoint union of an isolated vertex and an infinite path . the proof is complete . in the following theorem , we consider the prime index graph of cyclic groups . [ cyclic ] let @xmath138 , where @xmath102 are distinct primes and @xmath103 are positive integers . then @xmath139 . we know that @xmath140 . if @xmath2 and @xmath3 are two distinct subgroups of @xmath28 , then @xmath141 and @xmath142 , where @xmath143 for @xmath106 . so @xmath2 and @xmath3 are adjacent if and only if there exists an integer @xmath144 , @xmath145 , such that @xmath146 for @xmath147 and @xmath148 . thus @xmath149 and @xmath139 . [ regular ] let @xmath0 be a finite abelian group . if @xmath1 is regular , then @xmath150 and @xmath151 , where @xmath102 are distinct prime numbers . let @xmath152 , where @xmath102 are distinct primes and @xmath153 are positive integers . assume that @xmath154 , where @xmath155 is a positive integer and @xmath156 , for @xmath106 . we claim that @xmath157 for each @xmath107 , @xmath158 . by contradiction assume that @xmath159 , for some @xmath107 , @xmath158 . let @xmath160 be the number of subgroups of order @xmath105 in @xmath161 . obviously , the number of subgroups of order @xmath105 in two groups @xmath161 and @xmath162 are the same . hence by @xcite , we have @xmath163 . clearly , @xmath164 and @xmath165 . since @xmath1 is a regular graph , so @xmath166 . this implies that @xmath167 and hence @xmath157 , a contradiction . the claim is proved . thus @xmath0 is a cyclic group of order @xmath168 and by theorem [ cyclic ] , @xmath169 . now , since @xmath1 is a regular graph , @xmath170 and @xmath151 . let @xmath0 be a finite group . if @xmath1 is a @xmath171-regular graph , then @xmath172 and @xmath173 , where @xmath31 and @xmath96 are distinct primes . since @xmath174 , the order of @xmath0 has at most two distinct prime divisors . clearly , a @xmath31-group can not have exactly two subgroups of order @xmath31 . so assume that @xmath31 and @xmath96 are two distinct prime divisors of @xmath175 . suppose that @xmath2 and @xmath3 are subgroups of @xmath0 such that @xmath176 and @xmath177 . since @xmath174 , @xmath2 and @xmath3 are normal subgroups of @xmath0 . hence @xmath178 is a subgroup of @xmath0 and @xmath99 @xmath2 @xmath178 @xmath3 @xmath99 is a cycle in @xmath1 . now since @xmath1 is a @xmath171-regular graph , @xmath2 is a sylow @xmath31-subgroup and @xmath3 is a sylow @xmath96-subgroup of @xmath0 . thus @xmath179 and @xmath172 . in this section , we study those groups whose prime index graphs are connected . first we have the following lemma . it is clear that if @xmath0 is an infinite group , then there is no path between @xmath99 and @xmath0 in @xmath1 . so @xmath1 is not connected . if @xmath0 is an infinite simple group , by ( * ? ? ? * corollary 4.15 ) , @xmath0 can not have a proper subgroup of finite index . hence @xmath0 is an isolated vertex of @xmath1 . by @xcite , a finite group @xmath0 is supersolvable if and only if each subgroup of @xmath0 satisfies the converse of lagrange s theorem . so for finite supersolvable groups @xmath0 such as finite abelian groups and finite @xmath31-groups , @xmath1 is connected . ( note that every subgroup of @xmath0 is connected to @xmath99 . ) [ abeli ] let @xmath0 be a finite group and @xmath47 be a normal subgroup of @xmath0 . if @xmath48 is a connected graph and also for every subgroup @xmath180 of @xmath181 , @xmath182 is a connected graph , then @xmath1 is connected . assume that @xmath2 is a subgroup of @xmath0 . hence @xmath183 is a connected graph . since @xmath184 , so the graph @xmath185 is connected . this implies that there is a path between @xmath2 and @xmath186 in @xmath1 . now , since @xmath48 is connected , there is a path between @xmath186 and @xmath99 in @xmath48 . thus every subgroup of @xmath0 is connected to @xmath99 . therefore @xmath1 is connected . since @xmath0 is a solvable group , @xmath187 , for some positive integer @xmath11 . we prove the theorem by applying the induction on @xmath11 . if @xmath188 , then @xmath0 is an abelian group and so @xmath1 is a connected graph . assume that @xmath189 and @xmath187 . by the induction hypothesis , @xmath190 is connected . now , by theorem [ abeli ] , @xmath1 is a connected graph . if @xmath0 is a group of odd order , then @xmath0 is solvable ( feit - thompson theorem @xcite ) and by theorem [ solvable ] , @xmath1 is connected . moreover , suppose that @xmath191 , where @xmath56 and @xmath11 are positive integers with @xmath56 odd . if @xmath0 has a cyclic sylow @xmath171-subgroup , then by @xcite , @xmath0 has a normal subgroup of order @xmath56 and hence @xmath0 is a solvable group . thus @xmath1 is a connected graph . since every subgroup of a solvable group is solvable , by theorems [ abeli ] and [ solvable ] , we have the next result . first we prove that @xmath48 is a connected graph . let @xmath2 and @xmath3 be two distinct subgroups of @xmath47 . since @xmath1 is a connected graph , so there is a path @xmath2 @xmath192 @xmath193 @xmath194 @xmath3 from @xmath2 to @xmath3 in @xmath1 . we claim that by removing the same consecutive vertices in @xmath2 @xmath195 @xmath193 @xmath196 @xmath3 and keeping one of them we obtain a walk from @xmath2 to @xmath3 in @xmath48 . with no loss of generality , assume that @xmath197 and @xmath198=p$ ] , for some prime number @xmath31 . thus @xmath199 and we have @xmath200=\frac{\|l_{i+1}\cap n\|}{\|l_i\cap n\|}=\frac{\|l_in\|}{\|l_{i+1}n\|}\frac{\|l_{i+1}\|}{\|l_i\|}.\ ] ] hence @xmath201[l_{i+1}\cap n : l_i\cap n]=p$ ] . therefore @xmath202 or @xmath203=p$ ] . so the claim is proved . hence there is a path from @xmath2 to @xmath3 in @xmath48 which implies that @xmath48 is connected . + next , assume that @xmath2 and @xmath3 are two distinct subgroups of @xmath0 containing @xmath47 . suppose that @xmath2 @xmath193 @xmath194 @xmath3 is a path from @xmath2 to @xmath3 in @xmath1 . similar to the previous case , one can prove that @xmath180 @xmath193 @xmath205 @xmath206 is a walk from @xmath180 to @xmath206 in @xmath49 . thus @xmath49 is also a connected graph . * remark . * let @xmath11 be a positive integer . then @xmath210 is connected if and only if @xmath213 . also , @xmath211 is a connected graph if and only if @xmath213 . to prove the remark first assume that @xmath214 . hence @xmath29 is a solvable group and by theorem [ solvable ] , @xmath210 is a connected graph . if @xmath215 , we know that every proper subgroup of @xmath216 is solvable and @xmath216 contains a maximal subgroup of prime index , then @xmath217 is connected . also if @xmath213 , since @xmath29 is a normal subgroup of @xmath30 and @xmath210 is connected , by corollary [ sol ] , @xmath211 is connected . now , assume that @xmath218 . if @xmath11 is not a prime number , then by @xcite , @xmath29 has no subgroup of prime index and hence @xmath29 is an isolated vertex of @xmath210 . otherwise , if @xmath2 is a maximal subgroup of @xmath29 of prime index , then @xmath219 ( see @xcite ) . since @xmath72 is not a prime number , so @xmath210 is not connected . thus by theorem [ normal ] , @xmath211 is not connected . suppose that @xmath0 is the smallest group such that @xmath1 is not a connected graph . by theorem [ abeli ] , one can see that @xmath0 is a simple group . note that by the remark , @xmath222 is not a connected graph . on the other hand by @xcite , if @xmath0 is a non - abelian simple group of order at most 500 , then @xmath0 is isomorphic to one of the groups @xmath216 , psl@xmath223 , or @xmath212 . by remark , @xmath217 is connected . also by ( * theorem 6.26 ) , psl@xmath223 contains a maximal subgroup of index 7 and by @xcite , all subgroups of psl@xmath223 are solvable . hence @xmath224 is connected . thus @xmath221 . finally , for every non - abelian group @xmath0 with @xmath225 , since @xmath0 is not a simple group , so @xmath0 has a non - trivial proper normal subgroup @xmath47 . clearly , @xmath226 and @xmath227 are both less than 360 . thus by theorem [ abeli ] , @xmath1 is a connected graph .	let @xmath0 be a group . the prime index graph of @xmath0 , denoted by @xmath1 , is the graph whose vertex set is the set of all subgroups of @xmath0 and two distinct comparable vertices @xmath2 and @xmath3 are adjacent if and only if the index of @xmath2 in @xmath3 or the index of @xmath3 in @xmath2 is prime . in this paper , it is shown that for every group @xmath0 , @xmath1 is bipartite and the girth of @xmath1 is contained in the set @xmath4 . also we prove that if @xmath0 is a finite solvable group , then @xmath1 is connected .
in the middle of the last century , in the wake of observations made by navy officers during world war ii , according to which the radar echo of a plane flying near horizon above the ocean is modulated by interference fringes , australia was the home of the founding fathers of radio interferometry and the site of pioneering observations using the so - called `` sea - cliff interferometer '' @xcite . the principle of the method was to observe a radio source as it rises above the horizon with a single antenna located on top of a cliff above the ocean ; the direct wave and its reflection on the water surface interfere and produce interference fringes that allow for considerably improved angular resolution with respect to what was possible at that time . observations of solar spots @xcite , soon followed by observations of various radio sources @xcite , have then been reported . the present work is an illustration of the same mechanism causing correlations between apparent solar oscillations simultaneously detected by two distant observatories respectively located in ha noi ( viet nam ) and learmonth ( australia ) using radio telescopes operated at 1.415 ghz . in this case , oscillations are not observed on the rising sun but at large elevations : the reflected wave reaches the antenna in one of its side lobes , at large angle with respect to the beam . as a result , the oscillations have amplitudes of a few per mil , rarely exceeding 1% . they occur on the ground surrounding the antenna and their periods are in the range of a few minutes . the intriguing existence of correlations between the ha noi and learmonth observations had first been considered as an argument against a possible instrumental effect @xcite . it is now clearly established that the cause of the correlation is purely instrumental . the following sections develop this argument , a brief preliminary account of which has been presented elsewhere @xcite . interferences between the direct plane wave emitted by a radio source and detected in an antenna , and its specular reflection on a horizontal surface flat ground or ocean have been known to produce oscillations since the first days of radio interferometry @xcite . writing @xmath0 the difference in path length between the interfering waves , @xmath1 the frequency and @xmath2 the wavelength , here @xmath321 cm , the time difference between the two interfering waves is @xmath4 . introducing a parameter @xmath5 to account for the attenuation resulting from the reflection on ground and from the lesser gain of the side lobe in which the reflected wave is detected , the dependence on time @xmath6 of the detected signal reads @xmath7 in a time interval centred on @xmath8 ( where @xmath9 ) the time dependence of the oscillation reads therefore @xmath10\}\ ] ] describing the oscillation as a sine wave , @xmath11 $ ] , implies a period @xmath12 and a phase @xmath13 . in order to have @xmath14 positive , one prefers to write @xmath15 where the @xmath16 sign is for @xmath0 increasing with @xmath6 and the @xmath17 sign for @xmath0 decreasing with @xmath6 . relation 2 implies a fundamental relation between @xmath14 and @xmath18 @xmath19 which is characteristic of the multipathing effect . note that relation 2 allows for an evaluation of @xmath20 from measurements of @xmath14 and/or @xmath18 but not for a measurement of @xmath0 itself : the difference in path length can only be obtained up to an unknown integer number of wavelengths . equivalently , @xmath21 is only defined up to an unknown integer factor of @xmath22 . from figure 1 , @xmath23 where @xmath24 is the distance of the antenna to ground and @xmath25 the elevation of the source . as @xmath25 increases with @xmath6 in the morning and decreases with @xmath6 in the afternoon , so does @xmath0 . at a same time @xmath6 , from one day to the next , one expects therefore @xmath21 to increase in the morning in the northern hemisphere between december 22nd and june 21st . changing from morning to afternoon , or from northern to southern hemisphere or from december - june to june - december changes the sign of the daily variation of @xmath21 . if the hypothesis of specular reflection is slightly relaxed , reflections with an impact point farther away from the antenna than the specular impact ( i.e. reaching the dish at a smaller angle with respect to the main lobe ) are favoured with respect to reflections with an impact point closer than the specular impact . the reason is not only from pure geometry arguments but also because the reflected wave reaches the dish at large angle with respect to the beam and the antenna gain , on average , decreases with this angle . this causes the mean impact to be behind the specular impact and the path difference @xmath0 to take larger values than in the specular case ; as @xmath24 is proportional to @xmath0 , it is overestimated accordingly . the effect is easily simulated by describing the deviation from specular reflection by a gaussian dependence of the reflection probability on the angle @xmath26 of the reflected wave with respect to the specularly reflected wave , @xmath27 and the decrease of the antenna gain by another gaussian dependence on the angle @xmath28 between the reflected wave and the antenna axis , @xmath29 ( figure 1 ) . moreover , accounting for the irregularities of the reflecting surface can be approximated by a smearing of @xmath0 using a third gaussian , @xmath30 $ ] . and having image @xmath31 in the ground mirror . lower panel : departure from exact specular reflection ( mean ray ) , definition of the angles @xmath26 and @xmath28.,scaledwidth=38.0% ] an important consequence of the above mechanism is the existence of a correlation between the periods of simultaneous oscillations independently detected by two distant observatories such as ha noi and learmonth . consider two radio telescopes operated at a same frequency at heights @xmath32 and @xmath33 above a flat ground in observatories located at respective longitudes and latitudes @xmath34 and @xmath35 . writing @xmath36 , the periods read @xmath37 $ ] . in the approximation of a circular earth orbit , calling @xmath38 the sun declination and @xmath39 the hour angle @xmath40 ( @xmath41 when the sun crosses the meridian plane ) . hence , @xmath42\ ] ] with @xmath43 being a constant for each observatory . writing @xmath44 , @xmath45 and @xmath46 are trivially correlated , implying a correlation between the periods measured simultaneously by both observatories . if the two observatories are exactly at the same longitude , @xmath47 at any time . in practice , the longitudes of the two observatories should not be too different for them to have a chance to observe the sun simultaneously for long periods of time . in the ha noi / learmonth case , the two latitudes are nearly opposite ( @xmath48 ) and the two longitudes are nearly equal ( @xmath49 ) . in the approximation where @xmath50 , @xmath51 , @xmath52 rad ( observatory 1 being east of observatory 2 ) and writing @xmath53 the mean hour angle , @xmath54 values of 7 m and 6 m for learmonth and ha noi respectively.,title="fig:",scaledwidth=27.5% ] values of 7 m and 6 m for learmonth and ha noi respectively.,title="fig:",scaledwidth=33.0% ] around noon , the elevation is stationary and the period of the oscillations takes very large values outside the range where they can be detected . in the morning , @xmath55 , the elevation , and therefore the period of the oscillations increases with time . in the afternoon , @xmath56 , they decrease . changing @xmath57 into @xmath58 changes @xmath59 into @xmath60 . in the morning , both periods increase , that of observatory 1 faster than that of observatory 2 . in the afternoon , both decrease , that of observatory 2 faster than that of observatory 1 . this is schematically illustrated in the upper panel of figure 2 . the predicted correlation is essentially independent of the season during which observations are made : the only seasonal contribution is from the @xmath61 term which remains confined between 0.92 and 1 over the whole year and has therefore very little effect . when applied to a real observatory , the parameters that characterize the oscillations are strongly dependent on the environment of the antenna . in the cases studied here , ha noi and learmonth , the distance of the antenna to the reflecting surface is between 5 and 20 m , the latitude is @xmath62 and the wavelength is 21 cm . hence @xmath63 is between 1 and 5 min . the method used to detect oscillations requires their period to be of the order of a few minutes , say 2 to 10 minutes typically , excluding an interval of 1 to 2 hours around local noon , during which the period of the oscillations is too large to be observed . however , the rest of the time , one deals with a broad range of elevations , allowing the impact point on ground to reach close to the antenna . the learmonth antenna , 2.4 m ( 8 ft ) in diameter , ( figure 3 left ) is located in an airport on a pole erected some 7.5 m above a flat ground . the only nearby building is located south of the antenna , outside the range of azimuths toward which the antenna is pointing . on the contrary , the ha noi antenna , 2.6 m in diameter , ( figure 3 right ) is on top of a small building , some 20 m above ground and some 5 m above the flat roof , in a urban environment . one would therefore expect the multipath pattern observed at learmonth to be considerably simpler and easier to describe than that observed in ha noi . the data used in the present work cover the period october 24th 2013 to january 31st 2014 for both ha noi and learmonth . in addition learmonth data collected during the 10 central days of each month of the year have been included in the analysis . for both ha noi and learmonth , the signal to noise ratio is of the order of 500 . details of the procedure of data reduction are available in references 5 and 6 . after subtraction of solar flares and occasional interferences of human origin , the time dependence of the antenna temperature , averaged over the bandwidth , is searched for oscillations having periods in the 2 to 10 minutes interval . the method consists in evaluating , for each independent measurement , the amount of oscillation that the data can accommodate . each oscillation is evaluated over a 20 min interval centred on the particular measurement @xmath64 and is characterized by an amplitude @xmath65 , a phase @xmath66 , a period @xmath67 and a chi square describing the quality of the fit , @xmath68 . the significance of such oscillations can then be assessed objectively from the values of @xmath65 and @xmath68 without requiring human intervention . for each period @xmath14 and each measurement @xmath64 one calculates the mean value @xmath69 and the rms deviation with respect to the mean @xmath70 in a time interval centred on measurement @xmath64 and having a width @xmath14 . in a second step , a fit of the antenna temperature is performed over a 20 min time interval centred on measurement @xmath64 to a form @xmath71 $ ] . here @xmath6 spans the 20 min interval while @xmath72 is the time of measurement @xmath64 . parameters @xmath73 and @xmath21 measure the relative amplitude and phase of the oscillation respectively . the problem being linear in @xmath74 and @xmath75 allows for an easy explicit calculation of the parameters . figure 4 illustrates the procedure . significant oscillations are defined as having a small @xmath68 value , a large @xmath73 value ( meaning that the oscillation accounts for a major fraction of the signal fluctuation in the 20 min interval ) and a large @xmath76 value ( meaning that the amplitude of the oscillation exceeds noise level ) . different selection criteria have been tried and the robustness of the corresponding conclusions has been ascertained . distributions of time versus period using sensible selection criteria display very clear patterns that are illustrated in figure 5 and follow the same trend as expected from specular reflection on ground . examples of the evolution of the phase of the oscillations from one day to the next are displayed in figure 6 . here again , the dependence on season , hemisphere and time in the day is as expected from specular reflection on ground . and @xmath77 ( black ) . the lower panel compares data ( blue ) and fit ( red ) after subtraction of @xmath78 and division by @xmath79.,title="fig:",scaledwidth=35.0% ] and @xmath77 ( black ) . the lower panel compares data ( blue ) and fit ( red ) after subtraction of @xmath78 and division by @xmath79.,title="fig:",scaledwidth=32.5% ] m ( roof ) and @xmath80 m ( ground ) . lower panel : learmonth data ( red ) collected in the 10 central days of may 2012 . the blue lines are ground specular reflection multipath predictions for @xmath81 m.,title="fig:",scaledwidth=30.0% ] m ( roof ) and @xmath80 m ( ground ) . lower panel : learmonth data ( red ) collected in the 10 central days of may 2012 . the blue lines are ground specular reflection multipath predictions for @xmath81 m.,title="fig:",scaledwidth=30.0% ] having illustrated by a few examples in figures 5 and 6 the good qualitative agreement between the observed oscillations and the predictions of a multipath model assuming perfect specular reflection on ground , we now attempt a more quantitative analysis of the effect . the measurements of the period and phase of the oscillations provide independent evaluations of the altitude @xmath24 of the antenna above ground . from @xmath82 one obtains @xmath83 value).,scaledwidth=48.0% ] relation 6 relates the height @xmath24 of the antenna above the reflective surface to the measured value of the period of the oscillations under the hypothesis of specular reflection . figure 7 displays the distributions of @xmath24 obtained this way for the oscillations observed in learmonth and ha noi . the former is dominated by ground reflections while the latter displays a more complex structure revealing reflections from the observatory roof in the morning and late afternoon and from ground in the early afternoon . knowing @xmath24 , it is easy to map the impact coordinates on ground , @xmath84 and @xmath85 . they do not display any particular structure in the learmonth case , which is consistent with a flat ground , while revealing clearly distinct regions in the ha noi case , that are unambiguously associated with the topography of the environment as shown in figure 3 . using reasonable values for the parameters describing a small departure from exact specular reflection on ground , it is easy to obtain a good description of the @xmath24 distributions as illustrated in figure 7 where a common value of @xmath86 has been used for @xmath87 while @xmath88 takes values between @xmath89 and @xmath90 and @xmath91 . the quality of the data does not allow to measure these parameters precisely and other combinations of their values can be found that give also acceptable results . however , in all cases , the departure from exact specular reflection that can be accommodated remains small , not exceeding @xmath92 . an effect of the inclusion of such departure is to significantly lower the estimate of @xmath24 with respect to exact specular reflection : from 8.3 m to 7.7 m in the case of learmonth and , in the case of ha noi , from 6.2 m to 5.7 m for the roof and from 26 m to 21 m for ground , improving significantly the agreement with real dimensions ( respectively 7.5 m , 5.6 m and 17.7 m ) . distributions obtained from the learmonth ( upper panel ) and ha noi ( lower panel ) data in november - december 2013 . the blue lines show model predictions allowing for small departures from exact specular reflections ( see text ) . , title="fig:",scaledwidth=30.0% ] distributions obtained from the learmonth ( upper panel ) and ha noi ( lower panel ) data in november - december 2013 . the blue lines show model predictions allowing for small departures from exact specular reflections ( see text ) . , title="fig:",scaledwidth=30.0% ] the correlation expected between oscillations observed in the morning at learmonth from ground and in ha noi from reflections on the observatory roof is illustrated in figure 2 ( lower panel ) . the agreement with observation is remarkable given the crudeness of the model used . for oscillations having amplitudes in excess of @xmath93 for learmonth ( upper panel ) and ha noi ( lower panel ) data . the ha noi distributions display separately ground reflections ( black ) and roof reflections ( blue in the morning and red in the afternoon ) . the scale of ordinate is linear for learmonth and logarithmic for ha noi.,title="fig:",scaledwidth=30.0% ] for oscillations having amplitudes in excess of @xmath93 for learmonth ( upper panel ) and ha noi ( lower panel ) data . the ha noi distributions display separately ground reflections ( black ) and roof reflections ( blue in the morning and red in the afternoon ) . the scale of ordinate is linear for learmonth and logarithmic for ha noi.,title="fig:",scaledwidth=30.0% ] relation 3 , which relates independent measurements of the period and phase of the observed oscillations , offers a crucial test of their multipath nature . note that the path difference between the direct and reflected waves is of the order of magnitude of @xmath24 for the rather large values of the sun elevation associated with the observed oscillations , meaning several tens of wavelengths . as remarked earlier , the phase of the oscillation is measured up to an integer multiple of @xmath22 and it is not possible to measure @xmath24 directly from a single phase value but @xmath18 is simply evaluated from the phase difference between two successive measurements . the distribution of @xmath94 obtained this way is displayed in figure 8 for learmonth and ha noi oscillations having amplitudes in excess of @xmath93 , giving in both cases very strong evidence for the multipath nature of the effect . mean(rms ) values of 1.01(0.11 ) and 1.00(0.13 ) are obtained for learmonth and ha noi respectively , giving strong evidence for the multipath origin of the observed oscillations . note that ha noi data mix reflections from ground and from the observatory roof , spanning a broad range of @xmath24 values . the learmonth data cover the whole year and the simple topography implies reflections from a flat ground with a well defined @xmath24 value , 8.24 m on average . it fluctuates by only @xmath95 m over the year and the rms value of its distribution is 1.21 m on average . comparing the summer months ( november to february ) with the winter months ( may to august ) for the retained oscillations , the mean elevation of the sun is found to vary from @xmath96 to @xmath97 and the mean amplitude of the oscillations from @xmath98 to @xmath99 while the width of the distribution of @xmath94 remains constant to better than 10% of its value . the number of retained oscillations is nearly twice as large in winter than in summer . when observing the sun , multipath effects between the direct wave reaching the antenna in the main lobe and its reflection on ground reaching the antenna in a side lobe have been shown to produce correlations between the periods of oscillations observed independently by two distant radio telescopes . the case of observations made at 1.4 ghz in ha noi ( viet nam ) and learmonth ( australia ) has been studied in some detail . strong evidence for the multipath origin of the observed oscillations has been obtained from the relation between their periods and their phases . good agreement between observations and model predictions has been obtained and the departure from exact specular reflection that the data can accommodate has been shown to be small . the oscillations have periods and phases that are remarkably simple functions of time and are well described by the model . their amplitudes , at the level of a few per mil , are consistent with the gain drop expected between the main and side lobes of the antenna pattern . indeed , for a 65% aperture efficiency @xcite we expect a gain of some 30 dbi for the main lobe compared to some @xmath1720 dbi for a typical side lobe , namely a voltage ratio of some 3@xmath100 . the existence of a correlation between independent observations from two distant observatories , together with the large values of the elevation at which the oscillations were observed , had been used earlier @xcite as arguments against an instrumental explanation . it is now clear that the effect is of purely instrumental nature , making a search for genuine solar oscillations in this range of periods and amplitudes unfeasible with such instruments . we are deeply indebted to the learmonth solar observatory staff , who are making their data available to the public , and particularly to dr owen giersch for having kindly and patiently answered many of our questions related to such data and , in particular , for having first mentioned a possible contribution of multipathing . we are grateful to dr alain maestrini , dr pierre lesaffre and dr alan rogers and the anonymous referee for very useful comments . we acknowledge financial support from the vietnam national foundation for science and technology development ( nafosted ) under grant number 103.08 - 2012.34 , the institute for nuclear science and technology , the world laboratory , the odon vallet foundation and the rencontres du vietnam . bolton , j.g . 1948 , nature , 162 , 141 ; bolton , j.g . , & stanley , g.j . 1948 , nature , 161 , 312 ; bolton , j.g . , & stanley , g.j . 1948 , austral . res . , a1 , 58 ; bolton , j.g . , & stanley , g.j . 1949 , austral . res . , a2 , 139 ; bolton , j.g . , stanley , g.j . , & slee , o.b . 1949 , nature , 164 , 101 . bolton , j.g . 1982 , proc . astron . soc . australia , 4 , 349 . hiep , n.v . , 2013 , sol . 289 , 3 , 939 . mccready , l.l . , pawsey , j.l . , & payne - scott , r. 1947 , proc . a190 , 357 . phuong , n.t . , et al . 2014 , _ multipath generated correlations between apparent solar oscillations observed by two distant radio telescopes _ , submitted for publication in sol . phuong , n.t . , et al . 2014 , _ the vatly radio telescope : performance study _ , submitted for publication in comm . sullivan iii , w.t . 1991 , iau coll . 131 , asp conference series , vol . 19 , t.j . cornwell and r.a . perley ( eds . ) ;	a multipath mechanism similar to that used in australia sixty years ago by the sea - cliff interferometer is shown to generate correlations between the periods of oscillations observed by two distant radio telescopes pointed to the sun . the oscillations are the result of interferences between the direct wave detected in the main antenna lobe and its reflection on ground detected in a side lobe . a model is made of such oscillations in the case of two observatories located at equal longitudes and opposite tropical latitudes , respectively in ha noi ( viet nam ) and learmonth ( australia ) , where similar radio telescopes are operated at 1.4 ghz . simple specular reflection from ground is found to give a good description of the observed oscillations and to explain correlations that had been previously observed and for which no satisfactory interpretation , instrumental or other , had been found . radio detection multipath solar oscillations
the notion of entropy , @xmath1 lies at the heart of statistical mechanics where @xmath2 is the probability of being in state @xmath3 . it gives a measure of the number of microstates @xmath4 accessible to a system @xmath5 . it is also possible to define a relative entropy with respect to some reference set of configurations of probability @xmath6 ( such as those at the critical point or zero temperature ) , @xmath7 if @xmath8 represents parameters characterizing the class of models under consideration , the relative entropy , @xmath9 , induces a metric for two `` close '' configurations @xmath10 and @xmath11 : @xmath12 since @xmath13 and @xmath14 we find @xmath15 if we assume a gibbs - type distribution for the @xmath2 @xmath16 then the metric may be written as @xmath17 this metric is well known in statistics as the fisher - rao metric , and characterizes the `` closeness '' of probability distributions @xcite . for a one - parameter distribution the expectation value is the fisher information , @xmath18 , given by @xmath19 and is inversely related to the variance @xmath20 . in the multiparameter case this generalizes in the obvious manner to the covariance matrix . various authors @xcite have observed that the geometric invariants of the metric , in particular the curvature , @xmath21 , might be used to characterize the phase structure of statistical mechanical models . the first observation to note is that a non - interacting model , the ideal gas , has a flat geometry since @xmath22 . the suggestion was then that the curvature was a measure of the strength of interaction , an observation confirmed by a calculation in the case of a van der waals gas ( which is interacting ) : @xmath23 where @xmath24 is the inverse temperature , @xmath3 the pressure , and @xmath25 and @xmath26 are expressions appearing in the equation of state . this shows that the curvature diverges along the spinodal line where @xmath27 . these examples suggest that phase transitions are manifested as divergences in the scalar curvature associated with the fisher - rao metric , at least in two - parameter models . a natural question to ask is whether any of the standard scaling exponents associated with a transition can then be extracted from the behaviour of the curvature . this is best pursued by consideration of some specific examples . if we consider a spin model in field , the manifold of parameters , @xmath28 , is two - dimensional and parametrised by @xmath29 . in this case , the components of the fisher - rao metric take the particularly simple form @xmath30 where @xmath31 is the reduced free energy per site and @xmath32 . with the metric given in equ . ( [ frmetric ] ) , @xmath0 may be calculated succinctly as @xmath33 where @xmath34 . since the only scale present near criticality for a model displaying a higher - order transition is the correlation length , @xmath35 , it has been hypothesised on dimensional grounds that @xmath36 , where @xmath37 is the dimensionality of the system @xcite . if we assume that hyperscaling holds , @xmath38 , this leads to @xmath39 to test the behaviour of @xmath0 , one requires models which are solvable in field in order to obtain analytic expressions , and these are rather thin on the ground . indeed , @xmath0 has been calculated for the mean - field and bethe - lattice ising models @xcite and the above scaling behaviour verified . it has also been calculated for the one - dimensional ising model @xcite where it takes the remarkably simple form , @xmath40 in this case @xmath0 is positive definite and diverges only at the zero - temperature , zero - field `` critical point '' of the model . the correlation length is given by @xmath41 so that @xmath42 near criticality , and ( [ equscal ] ) holds there with @xmath43 as expected being zero @xcite . ] . given this rather short list of examples , any further additions would be very worthwhile in order to see which features are generic and which are particular to the models concerned . we expand the list incrementally here , looking at the _ one_-dimensional potts model , the _ two_-dimensional ising model coupled to two - dimensional quantum gravity and the _ three_-dimensional spherical model , which shares the critical exponents of the just mentioned coupled ising model . we also discuss the application of similar ideas to critical behaviour in various families of black holes . the present paper is essentially an amalgam of results presented in @xcite , with some added discussion of related work in black hole physics . the partition function for the 1d @xmath44-state potts model is given by @xmath45 , \label{zpotts}\ ] ] where the spins , @xmath46 , are defined on the sites , @xmath47 , of the lattice and where we have defined @xmath48 and @xmath49 for later calculational convenience . the model may be solved by transfer matrix methods @xcite , just as the 1d ising model . for general @xmath44 the full transfer matrix @xmath50 may be written as @xmath51 diagonal elements , @xmath52 , and a @xmath53 factor @xmath54 : @xmath55 the partition function is @xmath56 and the eigenvalues of @xmath57 are @xmath58 , where @xmath59 \ , ( y z ) ^ { - { 1 \over q } } \label{eigen}\ ] ] and @xmath60 . the reduced free energy per site in the thermodynamic limit , @xmath61 , is thus given by @xmath62 . it is straightforward to use this expression for the free energy in equ . ( [ equcurv ] ) to obtain the curvature , @xmath0 . in the current notation we re - derive the expression for the ising model between the ising and potts notations coming from the different spin definitions . ] as @xmath63 the expression for general @xmath44 is similar in form to this , and is @xmath64 where the coefficients @xmath65 and @xmath66 are smooth functions of @xmath67 and @xmath68 and do not diverge for finite ( physical ) temperature or field . furthermore @xmath69 ^ 2 + ( q- 1 ) 4 z . \label{eta}\ ] ] the expressions for @xmath70 and @xmath71 are very lengthy for general @xmath44 ( although still easily obtained ) so we do not reproduce them here . in zero - field ( @xmath72 ) the expression for @xmath0 is much more compact and is written for general @xmath44 as @xmath73 we see that as @xmath67 ranges from @xmath74 to @xmath75 , @xmath0 ranges from @xmath76 to @xmath75 . in particular , the sign of the @xmath77 ( @xmath78 ) limit of @xmath0 changes at @xmath79 , although the general morphology of @xmath0 as a function of @xmath67 and @xmath68 remains the same for all @xmath80 as we shall see below . the correlation length for the one - dimensional potts model is defined in a similar manner to that of the ising model , @xmath81 so @xmath82 for @xmath83 . we thus retrieve the scaling of @xmath0 for the one - dimensional potts model expected from equ . ( [ equscal ] ) , namely @xmath84 as @xmath85 . the exponents , as for the ising model , are @xmath43 . the general features of @xmath0 at non - zero temperature and field are perhaps easiest seen in a contour plot as a function of @xmath67 and @xmath68 . in fig . 1 we show the ising ( @xmath86 ) case which has certain non - generic features . the @xmath87 symmetry of the ising model appears as a @xmath88 symmetry in the plot of @xmath0 . in addition , one can see that @xmath0 is positive for all @xmath67 and @xmath68 . the maximum of @xmath0 for a given @xmath67 value lies along the zero field line at @xmath72 . in fig . 2 , @xmath0 is plotted for the @xmath89-state potts model , we see that there is no longer a @xmath88 symmetry and that @xmath0 is no longer positive definite . some years ago lee and yang @xcite addressed the question of how the singularities associated with field - driven phase transitions in ising - like spin models on lattices arose in the thermodynamic limit . this was later extended by various authors to other models and to temperature - driven transitions @xcite . lee and yang observed that the zeroes of the partition function for a spin model in a complex external field on a finite lattice would give rise to singularities in the free energy . in the thermodynamic limit these complex zeroes move in to pinch the real axis , signalling the onset of a physical phase transition . typically , the loci of zeroes are lines in the complex field or temperature plane and when the endpoints of such lines occur at non - physical ( i.e. complex ) external field values they can be considered as ordinary critical points with an associated edge critical exponent , usually dubbed the lee - yang edge exponent @xcite . the lee - yang zeroes for the one - dimensional potts model on a periodic chain with @xmath90 sites are given by the solutions @xcite of @xmath91 where @xmath92 are the eigenvalues given in equ . ( [ eigen ] ) and @xmath93 is odd . in the thermodynamic limit the locus of zeroes is determined by @xmath94 or @xmath95 ^ 2 + ( q- 1 ) 4 z = 0,\ ] ] which can be satisfied for complex ( in the @xmath86 ising case , purely imaginary ) values of @xmath96 . from these considerations , it is clear that @xmath0 will also diverge as the locus of zeroes is approached for both ising and potts models if we allow ourselves the liberty of an analytic continuation of the field to complex @xmath96 values once @xmath0 has been calculated , since @xmath97 and @xmath98 are finite as @xmath99 . the presence of the square root means that the divergence is characterised by an exponent @xmath100 which is the lee - yang edge exponent for the one - dimensional potts ( and ising ) model @xcite . the status of these observations is a little unclear to us , since the calculation of @xmath0 has assumed a real metric geometry throughout and such an arbitrary continuation in the final expression might be rather dangerous . it is nonetheless interesting that the lee - yang edge transition is still visible as a divergence in @xmath0 . the solution of the ising model on an ensemble of @xmath101 ( @xmath102-regular ) or @xmath103 ( @xmath89-regular ) planar random graphs was first presented by boulatov and kazakov @xcite , who were motivated by string - theoretic considerations , since the continuum limit of the theory represents matter coupled to 2d quantum gravity . they considered the partition function for the ising model on a single @xmath93 vertex planar graph with connectivity matrix @xmath104 : @xmath105 then summed it over all @xmath93 vertex graphs @xmath106 resulting in @xmath107 before finally forming the grand - canonical sum over graphs with different numbers @xmath93 of vertices @xmath108 where @xmath109 . this last expression could be calculated exactly as matrix integral over @xmath110 hermitian matrices , @xmath111 \right ) , \label{matint}\end{aligned}\ ] ] where the @xmath112 limit is to be taken to pick out the planar diagrams and the potential appropriate for @xmath101 ( 4-regular ) random graphs has been shown . when the matrix integral is carried out the solution is given in parametric form by @xmath113 where the function @xmath114 is @xmath115 \label{geq}\ ] ] and @xmath116 $ ] . in the thermodynamic limit the free energy per site is given by @xmath117 where @xmath118 is the appropriate low- or high - temperature solution of @xmath119 . when @xmath120 this may be solved in closed form , and although the solution is not available explicitly for non - zero @xmath96 it can still be developed as a power series in @xmath96 around the zero - field solutions in order to obtain expressions for quantities such as the energy , specific heat , magnetization and susceptibility . it was found that the critical exponents were given by @xmath121 , @xmath122 , @xmath123 , so the transition was _ third _ order with , intriguingly , the same exponents as the 3d spherical model on a regular lattice @xcite . if we carry out a perturbative expansion for the high - temperature solution , which is symmetric in @xmath96 and hence a series in even powers , we find @xmath124 where the coefficients in the series are most naturally expressed in terms of @xmath125 , as above . the expected scaling form of the various components of @xmath0 for a generic spin model in field is discussed at some length in @xcite , and we now recapitulate these results briefly for comparison with the specific results for the ising model on planar random graphs . the starting point is the scaling form of the free energy per spin near the critical point , @xmath126 where @xmath127 and @xmath128 are the scaling dimensions for the energy and spin operators . for @xmath129 , i.e. , in the unbroken high - temperature phase , we can use standard scaling assumptions to write this as @xmath130 where @xmath131 is a scaling function and we also define @xmath132 and @xmath133 for later convenience . in terms of the standard critical exponents @xmath134 and @xmath135 . this generic scaling form can now be substituted into equ . ( [ equcurv ] ) to find the scaling behaviour of the scalar curvature ( [ equcurv ] ) near criticality ( i.e. @xmath120 , @xmath136 ) , @xmath137 \over 2 a ( a -1 ) \psi_{+ } ( 0 ) } \epsilon^{-a}\ ] ] or , translating back to the standard critical exponents , @xmath138 the discussion in @xcite was intended to be as general as possible , one should note that for ising - like models with a @xmath87 symmetry all odd derivatives of the scaling function w.r.t . @xmath96 will vanish so @xmath139 rather than @xmath140 . this does not affect the stated scaling relations . however , one feature of these scaling relations does have an impact on the observed scaling for the ising model . generically one expects that @xmath141 , which contributes to both the metric and the determinant involved in calculating @xmath0 . if @xmath142 , i.e. @xmath143 , this naively suggests a vanishing @xmath144 at criticality , which will in general _ not _ be the case . there would instead be a contribution from a regular term , which would give a constant at the critical point . having such a constant term , which we take to be @xmath145 for notational convenience , modifies the scaling form of @xmath0 in the case @xmath146 , @xmath147 to @xmath148 or @xmath149 so the critical exponent @xmath3 no longer appears in the scaling exponent . by virtue of the boulatov and kazakov solution of the ising model on planar random graphs @xcite we can explicitly confirm these observations . since @xmath150 , we have @xmath151 , @xmath152 and the modified discussion of scaling should apply . we can now take the series expansion for @xmath153 from equ . ( [ equz0 ] ) , insert this into @xmath114 and use the resulting expression for @xmath31 in equ . ( [ equfl ] ) to calculate the various terms that appear in the scalar curvature @xmath0 as power series in @xmath154 . we find that the leading terms at @xmath120 , with @xmath155 and @xmath156 , and using @xmath157 as co - ordinates are @xmath158 a glance back at equ . ( [ equr3a ] ) shows that the modified scaling for @xmath142 that these incorporate is , indeed , followed . berlin and kac @xcite introduced the spherical model ( and the gaussian model ) in an attempt to understand how generic some of the features of onsager s solution @xcite of the @xmath159d ising model are for ferromagnetic spin models , particularly for other dimensions . in the spherical model , the @xmath160 condition on the value of the ising spins is relaxed , whilst retaining a global constraint on the _ total _ spin magnitude . with @xmath161 denoting the value of a spin at a site @xmath162 of a hypercubic lattice , the partition function is @xcite @xmath163 where @xmath90 is the total number of sites . this can be evaluated by exponentiating the constraint and using steepest descent , resulting in the following expression for the reduced free energy per site in the thermodynamic limit , @xmath164 : @xmath165 where @xmath166 the saddle - point value of @xmath68 which appears in the expression for the free energy in equ . ( [ sphsol ] ) is determined from @xmath167 the solution reveals no transition for @xmath168 and @xmath159 , and a transition with the exponents @xmath169 for @xmath170 . it is useful to note that , with @xmath120 , equ . ( [ saddle ] ) gives @xmath171 and hence @xmath172 ^ 3 } . \label{zscal2}\ ] ] the critical point is given by @xmath173 and @xmath120 @xcite , and the behaviour of @xmath114 in this region is determined by differentiating equ . ( [ saddlefunc ] ) twice and then expanding for the small @xmath174 values which give the dominant contribution . one finds @xmath175 a further differentiation gives @xmath176 and an integration yields @xmath177 where @xmath178 ^ 2 \approx 0.505\,462\,019\dots$ ] is the massless 3d lattice propagator at the origin , which can be expressed in terms of one of the three classical watson integrals @xcite and hence is given by the standard elliptic integral @xmath179 . this latter expression can be combined with equ . ( [ saddle ] ) with @xmath120 to give @xmath180 in which @xmath181 . equations ( [ gscal2 ] ) and ( [ gscal3 ] ) may then be substituted in equs . ( [ zscal1 ] ) and ( [ zscal2 ] ) to give the scaling of @xmath182 and @xmath183 , @xmath184 which we shall employ below in the calculation of the scalar curvature . we now move on to examine the scaling of the various terms contributing to @xmath0 in equ . ( [ equcurv ] ) for the spherical model . as we have remarked , the @xmath185 symmetry in the free energy per site , @xmath31 , of the spherical model means that any terms with an odd number of @xmath96 derivatives will automatically be zero when @xmath120 , hence @xmath186 . this leaves the non - zero terms in the scaling region @xmath187 we see that the expected general scaling of each term ( for @xmath143 ) does indeed apply and that overall we have , as in equ . ( [ equr3a ] ) , @xmath188 we thus see that calculating the scaling of @xmath0 for the @xmath89d spherical model for which @xmath121 gives results in accordance with expectations from general scaling arguments which take into account the negative @xmath3 , similarly to the ising model on planar random graphs . the thermodynamics of black holes has been a subject of abiding interest since the pioneering work of hawking @xcite and similar ideas to those discussed in the previous sections for statistical mechanical models have also been applied to investigations of the critical behaviour of various families of black hole solutions in general relativity . critical behaviour has arisen in several contexts in the study of black holes , ranging from the hawking - page @xcite phase transition in hot anti - de - sitter space and the pioneering work by davies @xcite on the thermodynamics of kerr - newman black holes , to the idea that the extremal limit of various black hole families might be regarded as a bona - fide critical point @xcite . it is the latter that is perhaps the closest to the work described here . in these studies the metrics used are the ruppeiner metric @xmath189 and the weinhold @xcite metric @xmath190 where @xmath191 is the entropy and @xmath192 the energy , with the @xmath193 being other extensive variables such as scalar charges . both metrics are related to the fisher - rao metric we have used by legendre transforms of the appropriate variables , and have also been employed in a statistical mechanical context . for example @xcite , a reissner - nordstrm black hole has a weinhold metric of the form @xmath194 where @xmath195 is the charge , and we can see that one of the metric components vanishes at @xmath196 . this led davies @xcite to suggest that there was a phase transition at this point , but transforming to the ruppeiner metric and choosing new co - ordinates gives a _ flat _ metric with @xmath197 everywhere . on the other hand kerr black holes which possess a spin @xmath198 display a curved ruppeiner geometry with a scalar curvature @xmath199 that diverges at the extremal limit @xmath200 it would thus appear that the general framework of looking for the signal of phase transitions in the geometric invariants of the information metric ( or related metrics ) can also be employed in these circumstances . indeed , the ads / cft correspondence identifies hawking - page type transitions with deconfinement in the dual gauge theories @xcite so it might be profitable to explore the information geometry view of such transitions further . d.j . and w.j . were partially supported by ec ihp network `` discrete random geometries : from solid state physics to quantum gravity '' _ hprn - ct-1999 - 000161_. d. brody and n. rivier , phys . e 51 * ( 1995 ) 1006 ; + d. brody and a. ritz , nucl . * b 522 * ( 1998 ) 588 ; + e.j . brody , phys . * 58 * ( 1987 ) 179 ; + d. brody and l. hughston , proc . london * a 455 * ( 1999 ) 1683 . j. lebowitz and o. penrose , comm . * 11 * ( 1968 ) 99 ; + g. baker , phys . 20 * ( 1968 ) 990 ; + r. abe , prog . * 37 * ( 1967 ) 1070 ; _ ibid . _ * 38 * ( 1967 ) 72 ; _ ibid . _ * 38 * ( 1967 ) 568 ; + s. ono , y. karaki , m. suzuki and c. kawabata , j. phys . japan * 25 * ( 1968 ) 54 ; + d. gaunt and g. baker , phys . rev . * b 1 * ( 1970 ) 1184 ; + p. kortman and r. griffiths , phys . * 27 * ( 1971 ) 1439 ; + m.e . fisher , phys . * 40 * ( 1978 ) 1611 .	although the notion of entropy lies at the core of statistical mechanics , it is not often used in statistical mechanical models to characterize phase transitions , a role more usually played by quantities such as various order parameters , specific heats or susceptibilities . the relative entropy induces a metric , the so - called information or fisher - rao metric , on the space of parameters and the geometrical invariants of this metric carry information about the phase structure of the model . in various models the scalar curvature , @xmath0 , of the information metric has been found to diverge at the phase transition point and a plausible scaling relation postulated . for spin models the necessity of calculating in non - zero field has limited analytic consideration to one - dimensional , mean - field and bethe lattice ising models . we report on previous papers in which we extended the list somewhat in the current note by considering the _ one_-dimensional potts model , the _ two_-dimensional ising model coupled to two - dimensional quantum gravity and the _ three_-dimensional spherical model . we note that similar ideas have been applied to elucidate possible critical behaviour in families of black hole solutions in _ four _ space - time dimensions .
the globular cluster system ( gcs ) of the galaxy has long been considered to contain `` fossil records '' on early dynamical and chemical histories of the galaxy ( e.g. , searle & zinn 1978 ; harris 1991 ; mackey & gilmore 2004 ) . therefore physical properties of the gcs , such as the abundance gradient ( e.g. , searle & zinn 1978 ) , radial density profile ( e.g. , zinn 1985 ; van den bergh 2000 ) , kinematics ( freeman 1993 ) , and proper motion ( dinescu et al . 1999 ) have been extensively discussed in the context of the galactic formation history . although recent numerical simulations of the galaxy formation based on the cold dark matter ( cdm ) model have provided a model explaining both dynamical and chemical properties of the galactic old stellar halo ( bekki & chiba 2000 , 2001 ) , theoretical models trying to explain the observed structural , kinematical , and chemical properties of the galactic gcs _ in a self - consistent manner _ have not yet been provided . this is partly because physical conditions required for globular cluster ( gc ) formation in low mass dwarfs , which can be the building blocks of the galaxy in the hierarchical clustering scenario , have not yet been completely clarified . recently , effects of reionization on galaxy formation have been extensively discussed , in particular , in the context of star formation histories and old stellar populations in low mass dwarfs ( e.g. , bullock et al 2000 ; gnedin 2000 ; susa & umemura 2004 ; grebel & gallagher 2004 ; bekki & chiba 2005 ) . these theoretical works suggested that suppression of star formation by heating and gas loss resulting from reionization can be a very important physical process for better understanding the observability and the early star formation activities in dwarfs , though grebel & gallagher ( 2004 ) did not find any clear fossil evidences that support the occurrence of such suppression at reionization era ( @xmath1 12.8 gyr ) in nearby dsphs . although it has already been discussed whether reionization can trigger ( cen 2001 ) or suppress ( santos 2003 ) the formation of globular clusters , it remains totally unclear how reionization determines dynamical and chemical properties of the galactic gcs . given the fact that unprecedentedly wealthy data sets on physical properties of the galactic gcs are now available ( e.g. , harris 1996 ; mackey & gilmore 2004 ) , it is worthwhile to discuss possible effects of reionization on the gc formation by comparing the observations with theoretical results in order to better understand the origin of the galactic gcs . the purpose of this letter is to demonstrate , for the first time , that physical properties of the galactic gcs can be significantly influenced by reionization by comparing observational data of the galactic gcs with numerical simulations including possible influences of reionization on gc formation . we particularly show how the radial profile of the number distribution of the galactic gcs can be influenced by _ the suppression of gc formation by reionization _ by using numerical simulations of the galaxy formation based on the cold dark matter ( cdm ) model . we firstly summarize three key observational results which can be interpreted as possible evidences for gc formation influenced by reionization in 2 , and then describe our numerical results in 3 and 4 . we suggest several observable gc properties that can be used to discuss reionization effects on gc formation for disk galaxies in 5 . interpretations given below for observational results would be somewhat speculative and are not unique , because there can be several alternative interpretations for each of the results . we however suggest that the origin of the results can be closely associated with some physical processes ( reionization etc . ) influencing the early galactic gc formation . it should be also stressed that the following observations could be only three among many of fossil evidences for the impact of reionization on the galactic gc formation . first is the steeper slope in the number density profile of the galactic gcs for the outer galactic halo ( van den bergh 2000 ) . as figure 1 reveals , showing the radial profile of the cumulative number distribution of the gcs ( @xmath2 ) , the gc distribution ( @xmath3 ) becomes progressively steeper toward the outer part of the galaxy . if @xmath3 is approximated as @xmath4 , where @xmath5 is the distance of a gc from the galactic center , @xmath6 is roughly @xmath7 for 2 @xmath8 @xmath5(kpc ) @xmath8 8 kpc ( van den bergh 2000 ) and @xmath9 for 20 @xmath8 @xmath5(kpc ) @xmath8 100 kpc ( it should be noted here that if @xmath2 is approximated as @xmath10 , then @xmath11 ) . the half - number radius of the galactic gcs ( @xmath12 ) can be estimated as 5.2 kpc . recent high - resolution cosmological simulations have shown that the cumulative radial distribution of dark matter subhalos in a milky way sized galaxy can increase linearly with @xmath13 , where @xmath14 corresponds to the virial radius of the halo , and have a half - number radius of @xmath1 @xmath15 ( see figure 2 in willman et al . 2004 ) . these simulation results mean that the radial number distribution @xmath16 of subhalos can be approximated as @xmath17 with @xmath18 @xmath19 @xmath20 and have the half - number radius ( @xmath21 ) of 209 kpc for @xmath14 = 348 kpc that is a reasonable value for the galactic mass of @xmath19 @xmath22 @xmath23 @xmath24 @xmath25 ( wilkinson & evans 1999 ) in the nfw profile ( model 5 in navarro et al . these observational and theoretical results raise the following question : if most of the galactic gcs originate from satellite galaxies in the subhalos ( e.g. , through accretion and tidal stripping of gcs ) , how can we explain the above remarkable differences both in the slope of density profiles in the galactic outer region ( @xmath6 and @xmath18 ) and in the half - number radius ( @xmath12 and @xmath21 ) between the gc distribution and the simulated subhalo one ? a possible explanation for this is that gc formation in outer subhalos ( i.e. , low mass dwarfs ) are strongly suppressed by some physical mechanisms . second is the apparently clear bimodality in the metallicity distribution of the gcs ( e.g. , zinn 1985 ; reed et al . theoretical studies suggested that the apparently clear secondary peak in the [ fe / h ] distribution can be due to secondary dissipative merger events that can form new , metal - rich gcs ( mrcs ) in the early dynamical history of the galaxy ( e.g. , ashaman & zepf 1992 ; bekki & chiba 2002 ; bekki et al . however recent semi - analytic study based on the cdm model ( beasley et al . 2002 ) has demonstrated that such clear gaps between the two peaks seen in galaxies can not be easily explained by such merger - induced gc formation _ without truncating the formation of metal - poor gcs at very high redshifts ( @xmath26 @xmath1 5)_. therefore the observed bimodality can possibly imply that the strong suppression of the formation of the galactic metal - poor gcs ( mpcs ) by some physical processes . third is the significant difference in gc number between the galaxy and m31 ( e.g. , van den bergh 2000 ) . although the total mass can be roughly the same between the galaxy and m31 ( evans et al . 2000 ) , the total number of gcs in the galaxy ( 160 @xmath27 20 ) is much smaller than that in m31 ( 400 @xmath27 55 ; van der bergh 2000 ) . this significant difference in the formation efficiency of gcs per galaxy mass can possibly indicate that gc formation is _ globally _ suppressed in the galaxy compared with m31 owing to some physical processes . thus we can interpret the above three observations as possibly implying the suppression of gc formation in the early galactic history . in the followings , we adopt a working hypothesis that reionization can be the mechanism responsible for the suppression of gc formation , and thereby investigate how it influences structural properties of the galactic gcs . we simulate the formation of a milky way sized galaxy halo in a @xmath28cdm universe with @xmath29 , @xmath30 , @xmath31 km @xmath32 @xmath33 , and @xmath34 , and thereby investigate merging / accretion histories of subhalos that can contain low mass dwarfs with gcs . the way to set up initial conditions for the numerical simulations is essentially the same as that adopted by katz & gunn ( 1991 ) and steinmetz & mller ( 1995 ) . we consider an isolated homogeneous , rigidly rotating sphere , on which small - scale fluctuations according to a cdm power spectrum are superimposed . the initial total mass , radius , spin parameter ( @xmath35 ) , and the initial overdensity @xmath36 of the sphere in the present model are @xmath37 , 30 kpc , 0.08 , and 0.26 , respectively . initial conditions similar to those adopted in the present study are demonstrated to be plausible and realistic for the formation of the galaxy ( e.g. , steinmetz & mller 1995 ; bekki & chiba 2001 ) . we start the collisionless simulation at @xmath38 ( = 30 ) and follow it till @xmath39 ( = 1 ) to identify virialized subhalos with the densities larger than @xmath40 , where @xmath41 is the critical density , at a given @xmath26 . the minimum number of particles within a virialized halo ( @xmath42 ) is set to be 32 corresponding to the mass resolution of 3.8 @xmath23 @xmath43 @xmath25 . for each individual virialized subhalo with the virialized redshift of @xmath44 , we estimate a radius ( @xmath45 ) within which 20 % of the total mass is included , and then the particles within @xmath45 are labeled as `` baryonic '' particles . 20 % of the _ outermost _ baryonic particles in a subhalo are labeled as `` gc particles '' so that the total number of gcs in the halo is 1 for the smallest halo with @xmath42 = 32 . thus the present simulation _ assumes _ that gcs can be formed in all low mass subhalos . in order to investigate the effects of the suppression of gc formation via reionization on the final structural properties of the simulated gcs , we adopt the following somewhat idealized assumption : _ if a subhalo is virialized after the completion of the reionization ( @xmath46 ) , gc formation is totally suppressed in the halo . _ therefore , `` gc particles '' in the subhalo with @xmath44 @xmath8 @xmath46 are not considered at all in the estimation of structural properties of the simulated gcs . recent wmap ( _ wilkinson microwave anisotropy probe _ ) observations have shown that plausible @xmath46 ranges from 11 to 30 ( spergel et al . 2003 ; kogut et al . 2003 ) whereas quasar absorption line studies can give the lower limit of 6.4 for @xmath46 ( fan et al . guided by these observations , we investigate the models with @xmath46 = 0 ( no reionization ) , 5 , 10 , and 15 . thus the present collisionless simulations follows the dynamical evolution of gcs originating from subhalos and do not include any _ dissipative _ formation of gcs associated with starbursts during merging / accretion of subgalactic halos . therefore , the simulated gcs can mimic old , mpcs rather than mrcs ( e.g. , `` disk gcs '' in the galaxy ) . all the calculations of the galaxy formation have been carried out on the grape board ( sugimoto et al . 1990 ) . total number of particles used in a simulation is 508686 and the gravitational softening length is 0.38 kpc . we consider that the final structure of the simulated gcs at @xmath39 ( = 1 ) is the same as that at @xmath26 = 0 , because the gcs is dynamically relaxed completely until @xmath39 for the present isolated model in which recent accretion / merging of dwarfs do not happen after @xmath39 . we used the cosmics ( cosmological initial conditions and microwave anisotropy codes ) , which is a package of fortran programs for generating gaussian random initial conditions for nonlinear structure formation simulations ( bertschinger 1995 ) . figure 2 shows how the galactic gcs can be formed through merging of subgalactic halos with gcs in the model with @xmath46 = 10 . subhalos grown from higher peaks of cdm density fluctuation are virialized earlier ( @xmath44 @xmath47 @xmath46 ) to contain older gcs within them , and these subhalos then merge with one another to form bigger subhalos with a larger number of gcs ( @xmath26 = 5.2 in figure 2 ) . numerous small subhalos without gcs can be seen around this redshift , which reflects the fact that about 47 % of all subhalos are formed later than @xmath46 from lower density peaks and thus suffer suppression of gc formation . these subhalos with and without gcs continue to merge ( @xmath26 = 3.0 ) to finally form a single massive halo ( @xmath26 = 2.1 ) with a number of bigger satellite halos with gcs . after accretion / merging of satellite subhalos and the resultant dynamical relaxation , a gcs finally forms with a relatively smooth gc density distribution , a triaxial shape , and the major axis of the distribution aligned with that of the dark matter halo . figure 3 shows the final radial profile of the cumulative number distribution ( @xmath2 ) of the simulated gcs in the model with @xmath46 = 0 , 10 , and 15 . it is clear form this figure that the profiles in the models with reionization are closer to the observed one than that in the model without reionization ( @xmath46 = 0 ) . in particular , the profile in the model with higher @xmath46 ( = 15 ) can more closely resemble the observed one than that in the model with lower @xmath46 ( = 10 ) . these results imply that the formation of the radial distribution of the galactic gcs could have been influenced by reionization . later accretion / merging of younger subhalos with gcs can add new gcs to the outer halo of the simulated galaxy and thus flatten the density profile of the gcs . in the model with @xmath46 = 15 , accretion / merging of younger subhalos can not influence the distribution of the gcs , because they do not contain gcs owing to later virialization with @xmath44 @xmath8 @xmath46 ( i.e. , suppression of gc formation ) . this is the essential reason why the model with @xmath46 = 15 can better reproduce the observation . the observed small half - number radius ( @xmath12 ) of 5.2 kpc ( e.g. , van den bergh 2000 ) can be also better reproduced by the model with @xmath46 = 15 showing @xmath12 of 5.0 kpc . the global mass profile of the simulated galactic dark matter halo does not depend on @xmath46 in the present study . therefore , figure 3 also suggests that the half - number radius of a gcs ( @xmath12 ) in a model with respect to the half - mass radius ( @xmath48 ) of the dark matter halo is smaller for higher @xmath46 . given the fact that ongoing observations on kinematics of gcss in galaxies enable us to infer mass profiles of the dark matter halos ( e.g. , bridges et al . 2003 ) , this @xmath12-@xmath48 ratio can be regarded as an important observable property that measures the influence of reionization on gc formation in galaxies . as expected from the adopted assumptions of the model , the total number of gcs ( @xmath49 ) depends strongly on @xmath46 in such a way that @xmath49 is larger for smaller @xmath46 ( e.g. , @xmath49 = 4986 and 116 for @xmath46 10 and 15 , respectively ) . this result might well provide a clue to a question as to why the galaxy has only @xmath8 200 gcs . thus the observed compact gc distribution with a radius - dependent slope of the distribution in the galaxy can be better explained by the model with @xmath46 = 15 . previous theoretical studies demonstrated that destruction of gcs is a key physical process that controls the radial profiles of gcss in galaxies ( e.g. , baumgardt 1998 ; fall & zhang 2001 ; vesperini et al . 2003 ) . in order to discuss how the simulated radial profiles of gcss can be modified by this destruction process , we numerically investigated the orbital evolution of gcs in the model with @xmath50 and thereby estimated the orbital eccentricity ( @xmath51 ) and pericenter ( @xmath52 ) of each of the gcs formed before reionization . we found that most of the gcs have highly radial orbits ( @xmath53 ) with smaller pericenter distances ( @xmath54 10kpc ) . aguilar et al . ( 1988 ) demonstrated that gcs with highly eccentric orbits ( @xmath55 ) and small pericenter distances ( @xmath54 2kpc ) can be completely destroyed by the galactic tidal field well within the hubble time . by combining this previous result with the above one on orbital properties of the simulated gcs , we found that about 55 % of the inner gcs in the simulation can be destroyed by the galactic tidal field . given the fact that selective destruction of gcs with higher @xmath51 and smaller @xmath52 can result in the flattening of the radial profiles of gcss ( e.g. , vesperini et al . 2003 ) , the above result implies that the radial density profile of the _ initial _ galactic gcs possibly composed mostly of gcs formed before @xmath46 can be significantly steeper than that observed now . furthermore the kinematics of the galactic gcs is observationally suggested to be fairly isotropic ( e.g. , freeman 1993 ) , which appears to be inconsistent of the simulated kinematics of the gcs composed mostly of gcs with highly eccentric orbits . this apparent inconsistency implies that the origin of the observed isotropic kinematics can be also closely associated with later destruction processes of gcs . the present study neglects the influence of the _ local _ uv field from the first - generation gcs ( and very massive stars ) formed in subgalactic clumps at @xmath56 in a protogalaxy on gc formation at @xmath57 in the protogalaxy . the simulated gcs at @xmath58 has a half - number radius of @xmath59kpc , and an order of magnitude estimation implies that gc formation within the half - number radius might well be influenced by local uv effects from population iii stars even before @xmath46 . accordingly structural and kinematical properties of the inner gcs , which can be formed well before @xmath46 , could be particularly modified by this local effect . more quantitative estimation of this local uv effects on gc formation can be done in our future numerical studies with more realistic and sophisticated models of gc formation . these future studies will also address a role of reionization in shaping the bimodal color distribution of the galactic gcs . bridges , t. , beasley , m. , faifer , f. , forbes , d. , forte , j. , gebhardt , k. , hanes , d. sharples , r. , zepf , s. 2003 , in extragalactic globular clusters and their host galaxies , 25th meeting of the iau , joint discussion 6 .	we propose that structural , kinematical , and chemical properties of the galactic globular clusters ( gcs ) can contain fossil information of the cosmic reionization history . we first summarize possible observational evidences for the influence of reionization on the galactic gc formation . we then show how structural properties of the gc system ( gcs ) in the galaxy can be influenced by suppression of gc formation due to reionization during the galaxy formation through hierarchical merging of subgalactic clumps , by using numerical simulations with and without suppression of gc formation by reionization . in particular , we show that if gc formation in dwarf galaxies that are building blocks of the galaxy and virialized after reionization era ( @xmath0 ) are completely suppressed , the present - day radial distribution of the galactic gcs depends strongly on @xmath0 . our numerical results imply that if gc formation after @xmath0 @xmath1 15 is strongly suppressed , the origin of the observed structural properties of the galactic gcs can be more naturally explained in the framework of the hierarchical clustering scenario .
the formation and evolutionary history of massive bulge - dominated galaxies has been a subject of considerable controversy over the past decade . on the one hand , the colours and detailed spectral energy distributions of these galaxies indicate that their stellar populations are predominantly old and metal - rich ( see renzini ( 2006 ) for a recent review ) . the dispersion in spectral properties between different systems is small , indicating that the stars in different galaxies were formed at roughly similar epochs and less than a few gigayears after the big bang ( e.g. bower , lucey & ellis 1992 ) on the other hand , a theoretical paradigm for the assembly of structure in the universe has emerged in recent years , which at first sight has appeared to contradict this observational data . according to this paradigm , the largest structures in the universe form _ hierarchically _ through the merging of small virialized condensations ( halos ) to form larger and larger systems . this means that the dark matter halos that host massive bulge - dominated galaxies are predicted to have assembled relatively recently . in addition , a significant fraction of these galaxies are expected to reside at the centers of massive halos , where gas is expected to reach high enough densities to be able to cool through radiative processes . the predicted cooling rates in these halos are high enough to produce a population of young stars that should be easily detectable in the spectra of their central galaxies ( see for example kauffmann , white & guiderdoni 1993 ) . in order to resolve this so - called `` cooling crisis '' , a number of recent models invoke feedback from the active galactic nuclei in massive galaxies . energy from the material accreting onto the central supermassive black hole is transported to the gas surrounding the galaxy either by jets that are known to exist in radio galaxies ( e.g. churazov et al 2001 ; ruszkowski et al 2004 ; best et al 2005 ; croton et al 2006 ) or by powerful outflows that are hypothesized to be triggered during galaxy - galaxy mergers ( granato et al 2004 ; di matteo , springel , hernquist 2005 ) . all these models are highly schematic , so it is valuable to use observational data to provide a more detailed understanding of the cooling , star formation and heating processes that regulate the evolution of massive galaxies . although the present - day star formation rates in massive galaxies galaxies are constrained to be low on average , there is observational evidence that stars do continue to form in a subset of these systems . the optical spectra of central cluster galaxies frequently exhibit line emission and blue continua ( mcnamara & oconnell 1989 ; cardiel , gorgas & aragon - salamanca 1998 ; crawford et al 1999 ) . excess ultraviolet light has also been detected in number of central cluster galaxies ( hicks & mushotzky 2005 ) . the inferred star formation rates in these objects are similar to rate at which gas is measured to be cooling at the very centers of the clusters using x - ray data from the chandra and xmm satellites ( mcnamara 2004 ) . recently , near and far - ultraviolet photometry from the galaxy evolution explorer ( galex ) has been assembled for a large sample of galaxies from the sloan digital sky survey . the advantage of studying galaxies at ultraviolet wavelengths is that their uv luminosities are very sensitive to low levels of residual star formation . the disadvantage is that rest - frame ultraviolet data can only be obtained from space for galaxies at low redshifts , so until recently the available galaxy samples have been small . ( 2005 ) and rich et al . ( 2005 ) presented analyses of the uv / optical color magnitude relation of two samples of early - type galaxies selected from the first matched galex / sdss sample . these studies found that the uv / optical colors of these galaxies exhibit significantly larger scatter than their optical colors . the main conclusion was that low - level residual star formation was common during the last billion years , even in bright early - type galaxies . rich et al . ( 2005 ) also noted that early - type galaxies with agn were much more likely than quiescent galaxies to show blue uv / optical colors . possible implications of the above results for models of elliptical galaxy formation have been considered in recent papers by kaviraj et al . ( 2005;2006 ) , martin et al ( 2006 ) and schawinski et al . ( 2006a , b ) . we note , however , that boselli et al ( 2005 ) carried out an analysis of uv properties of a sample of nearby elliptical galaxies in the virgo cluster , and found rather tight colour - magnitude relations for the objects in their sample . the ellipticals in this sample were selected using high quality photographic plates taken with the dupont telescope at las campanas observatory . the differences between the boselli et al analysis and the studies that use the sdss imaging data to select early - type galaxies , suggest that uv properties of early - type galaxies may be quite sensitive to the exact selection criteria used to define the different samples . in this paper , we adopt a different approach . using a parent sample of galaxies with galex detections that are matched to the sdss data release 4 ( dr4 ) sample , we have constructed a volume - limited sample of bulge - dominated galaxies in the local universe with near - uv and optical photometry , as well as optical spectroscopy from the sdss . unlike the previous studies , our sample is selected on the basis of bulge velocity dispersion and _ not _ on the basis of conventional hubble type . our sample is also limited to lie at redshifts less than 0.07 . at these distances the galaxies subtend a large enough solid angle on the sky to enable us to study the _ radial distributions _ of the young stars in the galaxies using both the galex and the sdss images . in addition , the spectra of these galaxies have very high signal - to - noise , enabling us to accurately deconvolve even intrinsically weak emission lines from the underlying stellar absorption continuum and to study the properties of agn in these objects . in section 2 , we describe the details of our sample selection . in section 3 , we compare the uv / optical colour relations of bulge - dominated galaxies with the `` traditional '' relations defined only in the optical . in section 4 , we analyze the spatial distribution of the young stars in galaxies with blue nuv-@xmath1 colours . in section 5 , we examine the link between low levels of ongoing star formation and agn activity in our sample . in section 6 , we study the stellar mass profiles of the galaxies in our sample . finally in section 7 , we summarize and interpret our results . since its launch in april , 2003 , galex has been conducting several surveys of the uv sky . details on the galex mission and surveys are given in martin et al ( 2005 ) . in this paper we make use of the galex medium deep survey ( mis ) . mis exposures are typically 1500 seconds long and reach a limiting magnitude @xmath3 . our data are taken from the galex mis ir1.1(+gr1 ) data release . the galex data include far - ultraviolet ( fuv ; @xmath4 , @xmath5 ) and near - ultraviolet ( nuv ; @xmath6 , @xmath7 ) images with a circular field of view with radius @xmath8 arcminutes . the spatial resolution is @xmath9 arcseconds . details of the galex satellite and data characteristics can be found in morrisey et al ( 2005 ) . the data were processed through the galex reduction pipeline at the california institute of technology . the pipeline reduces the data and automatically detects , measures , and produces catalogs of fuv and nuv fluxes for sources in the galex images . the galex catalogs were then matched to the sdss fourth data release ( dr4 ; adelman - mccarthy et al . 2006 ) spectroscopic sample . the matching procedure is described in seibert et al ( 2005 ) . matching to sdss provides a variety of photometric parameters in the @xmath10 photometric passbands as well as spectroscopic redshifts for a sample of 51,246 galaxies with uv detections . in this analysis , we exclude galaxies located more than 30 arcminutes from the mis field centers , because the photometry is not as reliable near the boundaries . in addition to the sdss photometric parameters , a large number of galaxy parameters derived from the sdss spectra are available in the value - added catalogs at the sdss website at the max planck institute ( http://www.mpa-garching.mpg.de/sdss ) . from these catalogs we use the stellar masses , stellar velocity dispersions , the spectral indices d@xmath11(4000 ) and h@xmath12 , the equivalent width of the h@xmath13 emission line ( eqw(h@xmath13 ) ) , and the agn classifiers . for detailed information about the derivation of these quantities , see kauffmann et al ( 2003a , c ) ; brinchmann et al ( 2004 ) and tremonti et al ( 2004 ) . throughout this paper we have assumed @xmath14 km s@xmath15 mpc@xmath15 , @xmath16 , and @xmath17 . the aim of the present analysis is to study a volume - limited sample of bulge - dominated galaxies . it is thus important to make sure that our sample , which is initially selected in the uv , is not biased _ against _ galaxies with no ongoing star formation . in figure 1 , we show how the nuv-@xmath1 and @xmath2 colours of galaxies depend on the fraction of the stellar mass that was formed in the past gyr . these curves were generated using the bruzual & charlot ( 2003 ) galaxev code assuming a solar metallicity stellar population and smooth , exponentially declining star formation histories with varying e - folding times . as can be seen from the plot , as the fraction of young stars in the galaxy goes to zero , the nuv-@xmath1 colours are predicted to saturate at a value of around 6.5 . the figure also shows that the nuv-@xmath1 colour remains sensitive to a much lower level of recent star formation than the optical @xmath2 colour . since our galex sample is limited at @xmath18 , figure 1 suggests that an @xmath1-band magnitude limit of 16.5 ( i.e. 1.2 magnitudes brighter than the limit of the main spectroscopic sample ) is required in order to ensure that our sample is not biased against galaxies with the very oldest stellar populations . the top left panel in figure 2 shows the nuv-@xmath1 colour distribution of galaxies with @xmath19 and nuv@xmath20 in our sample . we use sdss model magnitudes , which are well matched to the auto / kron magnitudes output by the galex pipeline . the magnitudes have also been corrected for foreground extinction using the values of e(b - v ) given by schlegel et al ( 1998 ) and the cardelli et al ( 1998 ) reddening curve . as can be seen , the colour distribution is clearly bimodal . the population of red galaxies is peaked at nuv-@xmath1 @xmath21 and there are very few galaxies with nuv-@xmath1@xmath22 . this gives us confidence that our sample is indeed complete in nuv-@xmath1 colour space . in the next two panels , we plot the stellar masses and the stellar velocity dispersions ( measured within the 3 arcsec fiber aperture ) of the red galaxies with nuv-@xmath1 @xmath23 as a function of their redshifts . these red objects are the most difficult to detect in the uv and they also have the highest @xmath1-band stellar mass - to - light ratios , so they determine the effective completeness limit of our sample . figure 2 shows that if we limit the sample to lie in the redshift range @xmath0 , we are complete down to @xmath24 and @xmath25 . we note that at @xmath26 , the sdss fibre aperture subtends a physical scale of only 3.2 kpc . the physical quantities that are measured from the spectra are thus relevant to the _ central bulge - dominated regions of the galaxy_. throughout this paper , we will interpret spectral quantities such as d@xmath11(4000 ) , h@xmath12 and eqw(h@xmath13 ) as measures of the age of the stellar population _ in the bulge_. the nuv-@xmath1 and @xmath2 colours , on the other hand , are sensitive to the age of the stellar population of the galaxy as a whole . in the bottom right panel of figure 2 we plot the stellar surface mass density @xmath27 as a function of stellar velocity dispersion for galaxies with @xmath28 , @xmath29 , @xmath19 , nuv@xmath30 and @xmath0 . this sample includes 1375 galaxies . the surface mass density @xmath27 is defined as @xmath31 , where @xmath32 is the half - light radius of the galaxy in @xmath33-band . as discussed in kauffmann et al ( 2003b ; 2006 ) , there is a sharp transition in the mean stellar age of galaxies at a stellar surface mass density @xmath27 of around @xmath34 kpc@xmath35 . galaxies with surface densities lower than this value have on average formed their stars at a constant rate for a hubble time ; galaxies with surface densities higher than this value have little or no ongoing star formation . figure 2 shows that essentially all galaxies in our sample defined at @xmath29 have surface densities greater than the transition value . the range of densities spanned by the objects in our sample is also very similar to that of elliptical galaxies ( bernardi et al 2003 ) and it is hence meaningful to refer to these objects as `` bulge - dominated '' systems . we also find that 85% of the galaxies in our sample have @xmath1-band concentration indices ( c = r90/r50 ) greater than 2.6 . strateva et al ( 2001 ) and shimasaku et al ( 2001 ) have studied how this index relates to `` by eye '' classification into hubble type . these authors find that c=2.6 marks a reasonably robust division between early - type galaxies ( e / s0/sa ) and the late - type spiral population . note that we choose not to apply further cuts to exclude the minority of later - type galaxies in our sample . the aim of the present paper is to examine possible evolutionary links between different galaxy populations at fixed bulge velocity dispersion or mass . it is thus critical to ensure that our sample is _ complete _ at a given value of @xmath36 . in figure 3 , we plot the nuv-@xmath1 and @xmath2 colours of the galaxies in our sample as a function of the stellar velocity dispersion measured within the sdss fiber . the points have been colour - coded according the emission line properties of the galaxies . galaxies with emission lines that are too weak to permit a secure classification are coloured in black , agn are coloured in red and star - forming galaxies are coloured in blue . as can be seen , the scatter in the nuv-@xmath1/@xmath36 relation is substantially larger than the scatter in the @xmath2/@xmath36 relation . the scatter increases at lower values of @xmath36 in both diagrams . it is striking that at velocity dispersions greater than @xmath37 125 km s@xmath15 , almost all the galaxies with blue nuv-@xmath1 colours are classified as agn . there are only a small number of galaxies with emission line ratios typical of star - forming galaxies and these all have small central velocity dispersions . these results are quantified in more detail in figure 4 , where we plot colour distributions in different ranges of stellar velocity dispersion . the black histograms represent the results for the full sample , while the red and blue histograms show the contributions from agn and star - forming galaxies respectively . figure 5 examines the relations between four different stellar population indicators : the nuv-@xmath1 colour , the @xmath2 color , the 4000 break index d@xmath11(4000 ) , and the equivalent width of the h@xmath13 emission line . as discussed in the previous section , the nuv-@xmath1 and @xmath2 colors are measures of the stellar populations of the galaxy as a whole , whereas d@xmath11(4000 ) and eqw(h@xmath13 ) probe the stellar populations within the central bulge - dominated region ( note that for galaxies classified as agn , the h@xmath13 will be sensitive to the ionizing flux from the central source as well as the hii regions in the galaxy . ) figure 5 shows that there is a tight , but strongly non - linear correlation between the @xmath2 and nuv-@xmath1 colours . as shown in figure 1 , galaxies with nuv-@xmath1 colours in the range 4 to 6 have formed less than 0.3 percent of their stars over the last gyr . the @xmath2 optical colours are completely insensitive to such low levels of star formation ; this is why @xmath2 is flat over this range in nuv-@xmath1 colour . the two spectral indicators measured within the fibre ( d@xmath11(4000 ) and eqw(h@xmath13 ) ) also exhibit a tight correlation . at fixed d@xmath11(4000 ) , agn ( red ) scatter to slightly higher h@xmath13 equivalent widths than star - forming galaxies ( blue ) , consistent with the notion that radiation from the active nucleus contributes to the ionization of the h@xmath13 line . nevertheless the relation between the two indicators is still quite tight for both classes of object . however , when we plot the fibre indicators as a function of the global nuv-@xmath1 colours , we see a huge scatter . galaxies with nuv-@xmath1 colours in the range @xmath38 have h@xmath13 equivalent widths that differ by more than an order of magnitude , and d@xmath11(4000 ) values that span the entire dynamic range of the index . the most obvious explanation for this enormous scatter is that the _ uv light does not trace the stellar population in the bulge . _ as we have noted previously , our sample is chosen to lie at redshifts below 0.07 and the physical aperture subtended by the sdss fibre is only a few kiloparsecs . the hypothesis that the uv light does not originate from the bulge is consistent with recent galex studies of very nearby galaxies , which indicate that a significant fraction of the uv light originates in the outer disks of galaxies ( thilker et al 2005 ) . popescu et al ( 2005 ) show that the uv - to - far infrared flux ratio in the galaxy m101 is a strong function of radius , with values monotonically decreasing from @xmath39 in the nuclear region to nearly zero at the edge of the optical disk . in the next section , we will prove that the uv light is biased towards the outer regions of the galaxies in our sample by analyzing their mean optical radial color profiles as a function of their global nuv-@xmath1 colours . the sdss photometric pipeline extracts azimuthally - averaged radial surface brightness profiles for all the objects in the survey . in the catalogues , this is given as the average surface brightness in a series of annuli with fixed angular dimension . results are stored in the @xmath40 and @xmath33-bands , so it is simple to compute radial colour gradients for the galaxies in our sample . in figure 9 , we plot @xmath41 as a function of physical radius for uv - bright ( upper panels ) and uv - faint ( lower panels ) galaxies with stellar velocity dispersions in the range @xmath42 . the average profile is shown as a solid line ; the dashed lines indicate the 10 - 90 percentiles in the range of colour spanned by the galaxies at a given radius . the colour profiles are smooth because we average over samples of between 100 and 300 objects . we only plot results out to radii where the photometric errors in the flux contained within the radial bin are less than @xmath43% . it is clear from figure 9 that uv - bright galaxies have very similar central colours to uv - faint galaxies , but are significantly bluer in their outer regions . uv - bright galaxies with strong emission lines in their central regions exhibit a larger scatter in their central @xmath41 colours , but their _ average _ nuclear colours are very similar to their uv - bright counterparts with weak emission lines . as was noted in the previous section , essentially all uv - bright galaxies with central stellar velocity dispersions in this range are classified as agn . in the bottom two panels , we compare the colour profiles of uv - faint galaxies that are classified as agn with those with emission lines that are too weak to classify . the colour profiles of the two kinds of galaxy are very similar . uv - faint agn are slightly bluer in their outer regions on average . in figures 6 and 7 , we show a montage of sdss @xmath44 and galex fuv , nuv colour images of bulge - dominated galaxies with blue nuv-@xmath1 colours ( nuv-@xmath45 ) . in figure 8 we show a sample of bulge - dominated galaxies with red nuv-@xmath1 colours ( nuv-@xmath46 ) for comparison . all the galaxies are selected to lie between redshifts 0.04 and 0.06 , and they all have central stellar velocity dispersions in the range @xmath47 . each postage stamp image is 100 arcsec on a side , which corresponds to a physical scale of @xmath37100 kpc . in figure 6 , we show galaxies with nuv-@xmath48 and low h@xmath13 equivalent widths measured within the fibre ( eqw h@xmath13 @xmath49 ) . figure 7 shows galaxies with nuv-@xmath50 and eqw h@xmath13 @xmath51 (note that essentially all of these are agn ) . there is no explicit emission line cut in figure 8 , but as shown in figure 5 , almost all galaxies with red nuv-@xmath1 colours have low emission line equivalent widths . the sdss colour images have been produced using an algorithm described in lupton et al ( 2004 ) . the algorithm uses an `` arcsinh stretch '' that allows faint , low - surface brightness features in a galaxy to be displayed , while simultaneously preserving the structure of the bright , high - surface brightness components . even at recessional velocity of 15,000 km / s , these galaxies display a rich complexity of structure in the sdss images , with outer rings , bars and spiral structure clearly visible in many objects . and we find that although outer disks are sometimes visible , it is no longer possible to discern star - forming regions in the images . ] the galex uv images have considerably lower spatial resolution than the sdss images , but they make it clear that the uv light is extended over large spatial scales . in figure 6 , there are galaxies where the uv appears to extend beyond the optical radius of the galaxy , out to galactic radii as large as 40 - 50 kpc . the uv - bright galaxies displayed in figure 7 have higher uv surface brightnesses than those in figure 6 , but the uv emission still clearly extends over the entire region traced by the optical light , and sometimes beyond . on the other hand the uv - faint galaxies displayed in figure 8 occasionally display some extended emission , but in most cases the uv light appears to be concentrated in the same bulge - dominated regions of the galaxies as the optical light . in section 3 , we showed that bulge - dominated galaxies with blue nuv-@xmath1 colours almost always have central emission line spectra characteristic of ionization by an active nucleus . in section 4 , we showed that the uv emission in bulge - dominated galaxies is a tracer of young stars in the _ region of the galaxy . this leads us to the following question : what , if any , is the physical connection between this extended uv emission and the activity in the nucleus of the galaxy ? in order to address this issue , it is necessary to move beyond simple classification of the galaxies in our sample into agn and and non - agn . heckman et al ( 2004 ) introduced the [ oiii ] emission line luminosity as an indicator of the rate at which matter is accreting onto the central supermassive black hole . there are two reasons why the [ oiii ] emission line luminosity is believed to be a good indicator of accretion rate : 1 ) [ oiii ] emission is relatively weak in metal - rich hii regions ; 2 ) in type i seyfert galaxies and quasars , the [ oiii ] line luminosity is well correlated with the continuum luminosity and by extension , the black hole accretion rate . assuming that the unified model is valid , the same should hold true in galaxies where the central engine is obscured . in follow - up work , heckman et al ( 2005 ) studied a sample of hard x - ray selected agn , and showed that the hard x - ray emission , which is believed to be an unbiased tracer of accretion on to the black hole , and the [ oiii ] emission - line luminosities were well - correlated over a range of 4 orders of magnitudes in luminosity in both type i and type ii agn . since we have measurements of the central stellar velocity dispersions for all the galaxies in our sample , we are able to estimate black hole masses using the relations given in tremaine et al ( 2002 ) . the ratio l[oiii]/m@xmath52 is then a measure of the accretion rate relative to the eddington rate . we note that the black hole masses and accretion rates estimated from the sdss spectra should be regarded as indicative rather than precise measures of these quantities . the stellar velocity dispersions in tremaine et al . ( 2002 ) were measured within phyical apertures that were on average a factor of 4 smaller than those corresponding to the 1.5 arsec fiber diameter for the galaxies in our sample . as discussed in heckman et al ( 2005 ) , there is at least a factor of a few uncertainty in the conversion from an [ oiii ] emission line luminosity to an accretion rate . nevertheless , as we will demonstrate , the stellar populations of the galaxies in our sample correlate remarkably tightly with l[oiii]/m@xmath52 . less than 15 % of the sdss sample is covered by galex observations . however , we have seen that even if we lack uv data , we can still use the sdss @xmath41 colour profiles to see whether there are young stars in the outer regions of our galaxies . we have thus chosen to stack galaxies with similar central velocity dispersions in bins of l[oiii]/m@xmath52 and to study how their optical colour profiles change as a function of black hole accretion rate . this is shown in figure 10 for galaxies with @xmath42 . solid cyan , blue , green , red , magenta and black curves show the average @xmath41 colour profiles of galaxies hosting agn with decreasing values of l[oiii]/m@xmath52 . throughout this paper . ] the dashed and dotted curves show the upper 90th and lower 10th percentiles of the distribution of @xmath41 colour at a given radius . the striking result indicated in this figure is that agn with higher accretion rates have bluer colours in their outer regions . in contrast , the average colour in the central regions of the galaxy appears to exhibit no dependence on the accretion rate onto the central black hole . at first sight , this might appear to be a paradoxical result . it would be difficult to come up with any physically motivated scenario that would predict the accretion onto the central black hole to be more strongly modulated by conditions in the outer rather than the inner regions of the galaxy ! it is important to remember , however , that galaxy colours are sensitive not only to the ages and metallicities of their stellar populations , but also to the amount of dust present in the interstellar medium . within the aperture covered by the sdss fibre aperture , information available from the sdss spectra allow us to probe mean stellar age and the amount of dust in an independent way . this is illustrated in figure 11 . in the upper two panels we plot the distribution of the two age - sensitive indices d@xmath11(4000 ) and h@xmath12 for galaxies in the same ranges in l[oiii]/m@xmath52 as shown in figure 10 . here we see a very clear progression in the mean value of these indices as a function of black hole accretion rate . the stellar populations in the central regions of galaxies with strongly accreting black holes are clearly younger . in the bottom panels we plot the distribution of @xmath53 , the attenuation of starlight due to dust in the @xmath33-band ( see kauffmann et al ( 2003a ) for more details ) , and the balmer decrement h@xmath13/h@xmath54 , which measures the extinction in regions of the galaxy with young , massive stars . both indicators show that there is more dust in the central regions of galaxies with strongly accreting black holes . we conclude that the reason why the average colours in the central regions of agn are so weakly correlated with accretion rate , is because the bluing that would be expected as a consequence of the decrease in the mean stellar age of the stars in the bulge is cancelled by the reddening due to the increase in the amount of dust . the _ exactness _ of this cancellation in the central bulge - dominated region of the galaxy is remarkable and is worthy of further consideration . in the outer ( presumably disk - dominated ) regions of the galaxy , age and dust effects no longer compensate each other exactly and as a result , these regions of the galaxy do become bluer as the age of the stellar population decreases . the results presented in figures 10 and 11 indicate that accretion rates onto central supermassive black holes respond to conditions in _ both _ the outer and the inner regions of the host galaxies . we now ask whether we can ascertain which response is the stronger of the two . to answer this question , we divide our sample into bins in outer @xmath41 colour ( measured at radii between 7 and 10 kpc ) and in inner stellar population age as measured by the h@xmath12 index . we study how the distribution in l[oiii]/m@xmath52 changes as a function of these measures of the age of the outer and the inner stellar populations . the results are shown in figures 12 and 13 . it is clear from these plots that the accretion rates respond much more strongly to the age of the inner stellar population . this conclusion is re - iterated in figure 14 , where we present the correlations between nuv-@xmath1 colour , d@xmath11(4000 ) and l[oiii]/m@xmath52 for galaxies with @xmath42 . as we have discussed , the nuv-@xmath1 colour is sensitive to the age of the stellar population in the outer galaxy ( effects due to dust are expected to be weak ) , d@xmath11(4000 ) measures the age of the stars in the inner bulge - dominated region , and l[oiii]/@xmath55 measures the accretion rate onto the black hole . as can be seen , the black hole accretion rate is much more tightly correlated with the mean stellar age in the bulge than in the outer galaxy . the black hole / outer galaxy and bulge / outer galaxy correlations both have a triangular shape . this triangular correlation tells us that galaxies with high black hole accretion rates and young bulges almost always have young stars in their outer regions . however , a young outer region _ does not guarantee _ that the bulge is young or the black hole accretion rate is high . the most straightforward interpretation of the results presented in the previous sections is that the uv light traces an extended reservoir of hi gas that surrounds a subset of the bulge - dominated galaxies in our sample . this outer gas is presumably in the form of a rotationally supported disk and it contains the fuel that is required for further growth of the the central supermassive black hole and the surrounding bulge . the triangular shape of the correlations between d@xmath11(4000 ) , l[oiii]/@xmath55 and nuv-@xmath1 colour presented in the first two panels of figure 11 implies that the presence of gas in this disk is a _ necessary , but not sufficient condition for further bulge and black hole growth . _ black holes with low accretion rates are found both in galaxies with old bulges and red nuv-@xmath1 colours ( these are galaxies that are nearly devoid of gas ) and in galaxies with old bulges and blue nuv-@xmath1 colours ( these galaxies do contain gas , but in the form of an extended disk ) . on the other hand , the tight correlation between l[oiii]/m@xmath52 and d@xmath11(4000 ) presented in the third panel of figure 14 implies that the presence of young stars ( and presumably gas ) in the bulge is a _ necessary and sufficient condition _ for further black hole growth . the strong correlation between black hole accretion rate and the amount of extinction measured within the sdss fiber aperture ( bottom panels of figure 11 ) is also clear evidence that black hole fuelling is strongly linked to the amount of cold gas in the inner region of the galaxy . we propose that strongly accreting black holes are only found in galaxies where the gas in the disk has become more centrally concentrated , perhaps as a result of angular momentum loss during an interaction . unfortunately , hi data is not available for the galaxies in our sample . however , we can test the hypothesis that the extended uv - bright components shown in figures 6 and 7 correspond to an outer disk by analyzing the radial distribution of the _ stars _ in these galaxies . to do this we need to transform the radial surface brightness profile of the galaxy into a stellar mass profile . bell & de jong ( 2001 ) used spiral galaxy evolution models to predict that the optical colours of galaxies are expected to be strongly correlated with their stellar mass - to - light ratios . both stellar age and dust attenuation affect the optical colours , but in such a way that @xmath56 can still be accurately predicted , provided that the recent star formation history of the galaxy has been reasonably smooth . we have estimated stellar mass - to - light ratios within the sdss fibre aperture using a method based on stellar absorption - line indices ( kauffmann et al 2003a ) . we have correlated our estimates of @xmath56 with a variety of optical colours and we find the tightest relation with the @xmath41 optical colour . this is shown in figure 15 . the black points show our sdss / galex sample of bulge - dominated galaxies , while the blue points are for galaxies with smaller stellar masses and velocity dispersions , but in the same redshift range as the galaxies in our bulge - dominated sample ( @xmath57 ) . note that we have taken care to plot the mass - to - light ratio in the @xmath58-band as a function of the @xmath41 fiber colour , so that all quantities are measured and evaluated within the same physical aperture . as can be seen @xmath56 is well - correlated with @xmath41 fiber colour over a range of more than a factor 10 in mass - to - light ratio and with an r.m.s . scatter of less than 0.1 dex . this gives us confidence that we can use the @xmath58-band surface brightness profile of a galaxy in conjunction with its @xmath41 radial colour profile to calculate the fraction of its stellar mass exterior to a given radius in a galaxy . the results of this calculation are shown in figure 16 . we stack galaxies within a narrow range in central stellar velocity disperion ( @xmath42 ) and study whether their stellar mass profiles differ according to whether they are uv - bright or uv - faint . in the left panel , solid black and red curves show average mass profiles for uv - faint galaxies , while solid blue and green curves are for uv - bright galaxies . the green curve shows the mass profile for uv - bright galaxies with weak central emission line strengths , while the blue curve is for uv - bright galaxies with strong central emission lines . ( note that the colour - coding is the same as in figure 9 ) . as can be seen , the stellar mass profiles of the uv - faint galaxies and the uv - bright galaxies with strong central emission line strengths are very similar . only uv - bright galaxies with weak central emission lines have significantly different mass profiles ; these galaxies have on average 10% more mass in their outer regions . in the right panel of figure 16 , we plot the stellar mass profiles of agn ordered according black hole accretion rate . the colour - coding is the same as in figure 10 . there is a trend for galaxies with higher black hole accretion rates to have more concentrated stellar mass profiles , but it is quite weak . only the galaxies with black holes that are accreting near the eddington rate appear significantly more concentrated . note that these strongly accreting systems also have total stellar masses that are about 10 - 20% smaller than the galaxies with weakly accreting black holes . the small difference in total stellar mass at fixed value of the central velocity dispersion @xmath36 is consistent with the notion that these galaxies all have very similar total masses , but that the more strongly accreting objects have slightly higher gas fractions . the evolutionary scenario suggested by the results presented in this section is one in which bulges and their central supermassive black holes form from gas located in an outer disk . these disks have presumably accreted from gas in the surrounding dark matter halo . so long as the galaxy remains undisturbed , very little of the gas from this disk will manage to reach the bulge and fuel an agn . these quiescent bulges / black holes with blue extended disks make up the left bottom corner of the triangular cloud of points shown in the first two panels of figure 14 . eventually , some process will lead to a flow of gas from the disk to the bulge . this triggers star formation in the bulge and significant growth of the central black hole . during this process , the increase of stellar mass in the bulge is larger than the increase in the disk , so the galaxy develops a more centrally concentrated stellar mass profile . once the gas in the disk is exhausted , the mass profile of the galaxy is indistinguishable from that of a classical elliptical galaxy . we have defined a sample of massive bulge - dominated galaxies with both galex photometry and spectroscopic data from the sdss . our choice of limiting @xmath1-band magnitude ensures that that the sample is complete in the near ultra - violet pass - band . we have also imposed stellar mass and redshift cuts to ensure that the sample is volume - limited for all galaxies above @xmath59 irrespective of their past star formation histories and present - day stellar mass - to - light ratios . we study galaxies with central velocity dispersions greater than 100 km / s . we have shown that these are bulge - dominated systems with stellar surface mass densities well above the value where galaxies transition from actively star forming to `` passive '' systems . the nuv data provide significantly higher sensitivity to low rates of star formation in bulge - dominated galaxies . our primary observational results can be summarized as follows : * bulge dominated galaxies exhibit a much larger spread in nuv-@xmath1 colour than in optical @xmath2 colour . the dispersion in colour decreases for galaxies with larger central velocity dispersions . * nearly all of the galaxies with blue nuv-@xmath1 colours are agn . * galex images and sdss color profiles demonstrate that the uv excess is associated primarily with an extended outer component of the galaxy . * when comparing fiber - based properties to global ones , we find `` triangular '' correlations . galaxies with red outer regions almost never have a strong agn or a young bulge . galaxies with blue outer regions have a wide range in bulge / black hole properties . galaxies with strongly accreting black holes and young bulges almost always have blue outer regions . * the black hole growth rate correlates much more strongly with the age of the stellar population in the bulge than in the outer region of the galaxy . * the amount of extinction in the bulge ( tracing the column density of the cold interstellar medium ) is also strongly linked to black hole growth and the age of the bulge stars . * at fixed central stellar velocity dispersion , we find that the radial distribution of the stellar mass in the host galaxy shows only small ( @xmath37 10% ) variations as a function of black hole growth rate and the colour of the disk . the suggested scenario is one in which the source of gas that builds the bulge and black hole is a relatively low mass reservoir of cold gas in the disk . the presence of the gas is necessary , but not sufficient for bulge and black hole growth . some mechanism must transport this gas inward in a time variable way . the duty cycle for such inflow periods ( once cold gas in the outer disk is present ) must be relatively large , because a significant fraction of the galaxies in our sample have both blue outer disks and star formation in the bulge . after the gas has been transported into the bulge , the relative timescales for bulge and black hole growth must be also similar , because there is a reasonably tight correlation between d4000 and l[oiii]/@xmath55 . the disk gas itself is likely to be the result of the accretion of gas from an external source . it is possible that these disks represent the repository of baryonic material that has managed to condense out of cooling flows in the centres of dark matter halos . as this gas is converted into stars in the bulge and disk , galaxies will turn red in nuv-@xmath1 . further inflow can bring them back into the blue nuv-@xmath1 sequence . one question that arises is whether the evolution of the disk occurs through secular processes , i.e. the required rearrangement of angular momentum and mass is achieved through collective phenomena such as bars , oval disks , spiral structure and triaxial dark matter halos ( see kormendy & kennicutt 2004 for a recent review ) , or whether the evolution is dominated by more violent and rapid processes , such as mergers and strong interactions . the similarity of the average stellar mass profiles of the quiescent and active galaxies implies that the processes that lead to episodic growth of the bulge and black hole do not usually involve major changes in the galaxy structure . we will address these issues in future work . in summary , the galex data presented in this paper have provided us with a new view of the bulge - dominated galaxy population in which star formation has largely shut down by the present day . we have learned that the transition from actively star - forming galaxies at low stellar surface mass densities to `` passive '' galaxies at high stellar surface densities does not imply that disk formation has ceased altogether , but that disks have become subdominant . in galaxies with large central velocity dispersions , the disks are inconspicuous in the optical . however , star - forming outer disks still dominate the energy output of the galaxy at ultraviolet wavelengths and this leads to a very different accounting of which galaxies in the local universe are truly `` red and dead '' . we have also seen that these uv disks contain the fuel for continued growth of central supermassive black holes . understanding the processes by which uv - bright disks form and why their formation is increasingly inhibited as the central velocity dispersion of the bulge increases will be an important next step in piecing together the puzzle of how massive galaxies came to be . galex ( galaxy evolution explorer ) is a nasa small explorer , launched in april 2003 . we gratefully acknowledge nasa s support for construction , operation , and science analysis for the galex mission , developed in cooperation with the centre national detudes spatiales ( cnes ) of france and the korean ministry of science and technology . funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s . department of energy , the japanese monbukagakusho , and the max planck society . the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions . the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the korean scientist group , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . adelman - mccarthy j. k. , et al . 2006 , , 162 , 38 bell e. f. , & de jong r. s. 2001 , , 550 , 212 bernardi , m. , et al . 2003 , , 125 , 1849 best p. n. , kauffmann g. , heckman t. m. , brinchmann j. , charlot s. , ivezi . , & white s. d. m. 2005 , , 362 , 25 boselli , a. et al . 2005 , , 629 , l29 bower r. g. , lucey j. r. , & ellis r. s. 1992 , , 254 , 601 brinchmann j. , charlot s. , white s. d. m. , tremonti c. , kauffmann g. , heckman t. , & brinkmann j. 2004 , , 351 , 1151 bruzual g. , & charlot s. 2003 , , 344 , 1000 cardelli j. a. , clayton g. c. , & mathis j. s. 1989 , , 345 , 245 cardiel n. , gorgas j. , & aragon - salamanca a. 1998 , , 298 , 977 churazov e. , brggen m. , kaiser c. r. , bhringer h. , & forman w. 2001 , , 554 , 261 crawford c. s. , allen s. w. , ebeling h. , edge a. c. , & fabian a. c. 1999 , , 306 , 857 croton d. j. et al . 2006 , , 365 , 11 di matteo t. , springel v. , & hernquist l. 2005 , nature , 433 , 604 granato g. l. , de zotti g. , silva l. , bressan a. , & danese l. 2004 , , 600 , 580 heckman t. m. , kauffmann g. , brinchmann j. , charlot s. , tremonti c. , & white s. d. m. 2004 , , 613 , 109 heckman t. m. , ptak a. , hornschemeier a. , & kauffmann g. 2005 , , 634 , 161 hicks a.k . , & mushotsky r. 2005 , , 635 , l9 kauffmann g. , white s. d. m. , & guiderdoni b. 1993 , , 264 , 201 kauffmann g. et al . 2003a , , 341 , 33 kauffmann g. et al . 2003b , , 341 , 54 kauffmann g. et al . 2003c , , 346 , 1055 kauffmann g. , heckman t. m. , de lucia g. , brinchmann j. , charlot s. , tremonti c. , white s. d. m. , & brinkmann j. 2006 , , 367 , 1394 kaviraj s. , devriendt j. e. g. , ferreras i. , yi s. k. , 2005 , , 360 , 60 kaviraj s. et al . 2006 , , in press(astro - ph/0601029 ) kormendy j. , & kennicutt r. c. , jr . 2004 , , 42 , 603 lupton r. , blanton m. r. , fekete g. , hogg d. w. , omullane w. , szalay a. , & wherry n. 2004 , pasp , 116 , 133 martin d. c. et al . 2005 , , 619 , l1 martin d. c. et al.,2006 , , submitted mcnamara b. r. , & oconnell r. w. 1989 , , 98 , 2018 mcnamara b. r. 2004 , rcfg.proc , 177 morrissey p. et al . 2005 , , 619 , l7 popescu c. c. et al.,2005 , , 619 , l75 renzini a. 2006 , , 44 ( astro - ph/0603479 ) rich r. m. et al . 2005 , , 619 , l107 ruszkowski m. , brggen m. , & begelman m. c. 2004 , , 611 , 158 schawinski , k . 2006a , , in press(astro - ph/0601036 ) schawinski k. et al . 2006b , nature , in press seibert m. et al . 2005 , , 619 , l23 schlegel d. j. , finkbeiner d. p. , & davis m. 1998 , , 500 , 525 shimasaki , k. et al . 2001 , , 122 , 1238 strateva , i. et al . 2001 , , 122 , 1861 thilker d. a. et al . 2005 , , 619 , l79 tremaine s. et al . 2002 , , 574 , 740 tremonti c. a. et al . 2004 , , 613 , 898 yi s. k. et al . 2005 , , 619 , l111	we analyze a volume - limited sample of massive bulge - dominated galaxies with data from both the sloan digital sky survey and the galaxy evolution explorer ( galex ) satellite . the galaxies have central velocity dispersions greater than 100 km / s and stellar surface mass densities that lie above the value where galaxies transition from actively star forming to passive systems . the sample is limited to redshifts @xmath0 . at these distances , the sdss spectra sample the light from the bulge - dominated central regions of the galaxies . the galex nuv data provide high sensitivity to low rates of _ global _ star formation in these systems . our sample of bulge - dominated galaxies exhibits a much larger dispersion in nuv-@xmath1 colour than in optical @xmath2 colour . the dispersion increases for galaxies with smaller central velocity dispersions and nearly all of the galaxies with bluer nuv-@xmath1 colours are agn . both galex images and sdss colour profiles demonstrate that the excess uv light is nearly always associated with an extended disk . when comparing fibre - based estimates of stellar age to global ones , we find that galaxies with red outer regions almost never have a young bulge or a strong agn . galaxies with blue outer regions have bulges and black holes that span a wide range in age and accretion rate . galaxies with young bulges and strongly accreting black holes almost _ always _ have blue outer disks . the black hole growth rate correlates much more strongly with the age of the stars in the bulge than in the disk . our suggested scenario is one in which the source of gas that builds the bulge and black hole is a low mass reservoir of cold gas in the disk . the presence of this gas is a necessary , but not sufficient condition for bulge and black hole growth . some mechanism must transport this gas inwards in a time variable way . the disk gas itself is likely to be the result of the accretion of gas from an external source . as the gas in the disk is converted into stars , galaxies will turn red , but further inflow can bring them back into the blue nuv-@xmath1 sequence .
the phenomenon of anderson localization @xcite appears in a variety of quantum physical systems including electron transport in disordered solids and waves in random media ( see e.g. @xcite ) . it is usually analyzed in the frame of hermitian or unitary matrices . recently , the localization properties of nonunitary complex matrices has been analyzed for euclidean matrices @xcite in relation to light and wave localization @xcite . in this work we analyze the possibilities of anderson like localization and delocalization for the matrices belonging to the class of markov chains and google matrix @xmath0 @xcite . such matrices have real nonnegative elements with the sum of elements in each column being equal to unity . for a directed network one first defines an adjacency matrix @xmath1 which has element @xmath2 if a node @xmath3 have a link pointing to node @xmath4 and zero otherwise . the columns with only zero elements ( _ dangling nodes _ ) are replaced by columns with @xmath5 where @xmath6 is the matrix size . the elements of other columns are renormalized in such a way that their sum becomes equal to unity ( @xmath7 , @xmath8 ) . thus we obtain the matrix @xmath9 of markov transitions . then the google matrix @xmath0 of the network takes the form @xcite : @xmath10 here , the damping factor @xmath11 is taken in the range @xmath12 . in the context of the world wide web ( www ) the term @xmath13 describes for a random surfer a probability to jump on any node of the network . the above construction of @xmath0 has been proposed by brin and page @xcite to describe the structure of the world wide web ( www ) . for the www it is assumed that the google search engine uses @xmath14 @xcite . we can also consider a generalized case of weighted markov transitions @xmath9 corresponding to real positive elements of @xmath1 like happens for the world trade network ( see e.g. @xcite ) . the matrix @xmath0 belongs to the class of perron - frobenius operators , its largest eigenvalue is @xmath15 and other eigenvalues have @xmath16 @xcite . the right eigenvector at @xmath15 , which is called the pagerank ( @xmath17 ) , has real nonnegative elements @xmath18 and gives a probability @xmath18 to find a random surfer at site @xmath4 . it is possible to rank all nodes in a decreasing order of pagerank probability @xmath19 so that the pagerank index @xmath20 counts all @xmath6 nodes @xmath4 according their ranking , placing the most popular nodes at the top values @xmath21 . usually for many real directed networks the distributions of number of ingoing and outgoing links are described by a power law ( see e.g. @xcite ) , generating an average approximately algebraic decay of pagerank probability @xmath22 with @xmath23 . some examples of directed networks can be found in @xcite . it is important to note that matrices of google class practically have not been studied in physical systems even if they naturally appear in the frame of ulam networks generated by the ulam method for dynamical maps in a coarse - grained phase space ( see e.g. @xcite ) . therefore , it is interesting to see if the phenomena of anderson localization and anderson delocalization transition can appear for google matrices . certain indications on a possible anderson transition for the ulam networks , built from dissipative maps , have been reported in @xcite with more detailed discussions presented in @xcite . thus , it would be useful to find random matrix models which are able to reproduce typical properties of spectrum and pagerank decay in real directed networks . however , the results presented in @xcite show that the full matrix @xmath0 with random matrix elements have an unrealistic spectrum and hence other random matrix models of @xmath0 should be developed . the models discussed in @xcite give certain indications of delocalization of eigenstates of @xmath0 but the spectrum of @xmath0 in these models has a large gap and is very far from the spectra of real directed networks . with this aim we describe below a number of random google matrix models and analyze the properties of their spectra and eigenstates . we use certain spectral properties of small size _ orthostochastic _ matrices with @xmath24 established in @xcite . we start from a description of various random matrix models of google matrix @xmath0 presenting the results of their spectral properties in next section . following @xcite we consider _ orthostochastic _ matrix blocks @xmath25 of size @xmath26 . orthostochastic _ property means that @xmath27 , where an orthogonal matrix @xmath28 has random matrix elements obtained via random rotations . since @xmath28 is an orthogonal matrix the matrix @xmath29 is bistochastic with @xmath30 @xcite . the main reason to use such blocks @xmath29 is a similarity of complex spectrum of random matrix ensemble of @xmath29 with the spectrum of @xmath0 of university networks of cambridge and oxford , as discussed in @xcite . the size @xmath31 can be considered as preferential random links between a group of @xmath32 friends . however , a weak point of the random ensemble of @xmath29 @xcite is a small matrix size @xmath33 , while for the above universities we have @xmath34 . google matrix eigenvalues @xmath35 ( a , b ) , and ipr @xmath36 of eigenvectors as a function of @xmath37 ( c , d ) . panels show data for rmz3 model ( sec.2.1 ) at fixed amplitudes @xmath38 ( a , c ) and for random amplitudes @xmath39 ( b , d ) . the green circle shows @xmath40 ( a , b ) ; the green horizontal line shows maximal possible @xmath41 ( c , d ) ; the gray band in ( a , c ) highlights specific states ( see text ) . here the total number of nodes is @xmath6=8000 . ] to go to large values of @xmath6 in matrix @xmath9 we construct the * random matrix model @xmath42 ( rmz3 ) * as follows : we place blocks @xmath29 of size @xmath43 on the main diagonal with weights @xmath44 and on two adjacent upper and lower diagonals with weight @xmath45 , where @xmath46 @xmath47 are random numbers uniformly distributed in some interval @xmath48 ; each block represents a random realization of @xmath29 ; then the matrix @xmath0 of total size @xmath6 is built from @xmath49 via the equation ( [ eq1 ] ) . here we consider two cases with a constant @xmath50 and the interval range @xmath51 ( see fig . [ fig1 ] ) . obviously , by construction the final matrix belongs to the google matrix class . we use notations @xmath52 and @xmath53 for the matrices @xmath49 and @xmath0 of this model . the model * rmz3s * is obtained from rmz3 by adding shortcut links between blocks @xmath29 in the upper triangle of the whole matrix @xmath49 , the blocks of shortcut links are placed randomly in this part of @xmath49 . the amplitude of transitions from one block to another block ( outside of three - diagonal blocks of rmz3 ) is taken at some fixed value @xmath54 . the shortcut blocks are randomly and uniformly distributed over the upper triangle of the whole matrix . _ after _ adding the shortcut blocks the columns affected by shortcut blocks are renormalized to unity . in this way the obtained matrix @xmath49 again belongs to the google matrix class . the blocks of shortcuts are placed randomly in the upper triangle of matrix @xmath49 , their number @xmath55 is determined by the parameter @xmath56 . in fact @xmath57 gives the ratio of shortcut blocks to the number of blocks @xmath58 in the model rmz3 . again each block @xmath29 in the main three - diagonal part of rmz3 and blocks at shortcut positions are taken as random and independent realizations for each block . we note that the shortcuts between single nodes have been used for studies of quantum chaos and anderson transition in the small world anderson model ( see @xcite ) . the results for rmz3s model are shown in figs . [ fig2 ] , [ fig3 ] . same as in fig . [ fig1 ] for rmz3s model ( sec.2.2 ) with @xmath51 , shortcut amplitude @xmath59 , @xmath60 ( a , c ) and @xmath61 ( b , d ) . here @xmath62 . ] dependence of @xmath63 for models rmz3s ( sec.2.2 ) in panels ( a , b ) and rmz3f ( sec.2.3 ) in panels ( c , d ) ; here @xmath51 . in panel ( a ) we have @xmath60 ( black symbols ) and @xmath61 ( blue symbols ) at @xmath62 ; the fitted algebraic dependence is shown by straight dashed lines with parameters : @xmath64 , @xmath65 at @xmath60 and @xmath66 , @xmath67 at @xmath61 ; panel ( b ) shows the dependence of @xmath68 on @xmath57 with the full curve for @xmath62 and triangles , crosses and circles for @xmath69 respectively ; the amplitude of shortcut elements is @xmath59 . in panel ( c ) we have @xmath70 ( black symbols ) and @xmath71 ( blue symbols ) at @xmath72 and @xmath62 ; the fitted algebraic dependence is shown by straight dashed lines with parameters : @xmath73 , @xmath74 at @xmath75 and @xmath76 , @xmath77 at @xmath78 ; panel ( d ) shows the dependence of @xmath68 on @xmath57 for @xmath79 ( blue ) and @xmath80 ( red ) with the full curves for @xmath62 and triangles , crosses and circles for @xmath69 respectively . ] the results obtained in @xcite show that a triangular matrix of google matrix class has a tendency to have a realistic pagerank probability decay with @xmath81 and have some eigenvalues of finite amplitudes @xmath82 . due to these indications we construct a matrix @xmath83 in the following way : @xmath84 random numbers @xmath85 from the interval @xmath86 are placed on random positions of the upper triangle of matrix of size @xmath6 , then all columns are renormalized to unity and columns with all zero elements are replaced by columns with all elements @xmath5 . then we construct the matrix @xmath0 of the * model rmzf * as : @xmath87 here @xmath88 determines a measure of contribution of @xmath83 with @xmath89 . the number of nonzero random elements @xmath84 is given by parameter @xmath90 . the results for the rmzf model are shown in figs . [ fig3],[fig4 ] , [ fig5 ] . spectrum ( a , b ) and ipr @xmath36 dependence of @xmath37 for the model rmz3f ( sec.2.3 ) at @xmath91 ( a , c ) and @xmath92 ( b , d ) ; here @xmath51 , @xmath79 , @xmath62 ; circle and horizontal lines are as in fig . [ fig1 ] . ] panel ( a ) shows dependence of maximal ipr @xmath36 ( for states with @xmath93 ) on parameter @xmath57 for the model rmz3f ( sec.2.3 ) at @xmath62 . dependence of maximal ipr @xmath36 on @xmath6 is shown in panels ( b ) at @xmath78 and ( c ) at @xmath60 ; error bars show statistical error , if it is larger than symbol size , obtained from @xmath94 disorder realizations . we use @xmath95 at @xmath96 , @xmath97 at @xmath98 , @xmath99 at @xmath62 , @xmath100 at @xmath101 . in all panels @xmath102 ( circles ) and @xmath103 ( triangles ) , @xmath51 ; the straight green lines show dependence @xmath41 ; the straight gray lines in ( b , c ) show the fitted dependence ( see text ) . ] we use the usual anderson model @xcite with diagonal disorder terms @xmath104 and transitions @xmath105 to nearby sites on a lattice in dimension @xmath106 : @xmath107 where indexes in bold are vectors in @xmath106-dimensional space . on the basis of ( [ eq3 ] ) we construct the matrices @xmath49 and @xmath0 . thus we consider the dimensions @xmath108 corresponding to square and cubic lattices . the matrix @xmath49 is constructed as follows : each transition matrix element , corresponding to @xmath105 terms , in the anderson model in dimension @xmath106 ( [ eq3 ] ) is replaced by a random number @xmath46 uniformly distributed in the interval @xmath109 $ ] , the diagonal element @xmath110 is replaced by unity minus the sum of all @xmath46 over @xmath111 nearby sites ( @xmath112 ) . the asymmetric matrix @xmath49 constructed in this way belongs to the google matrix class . thus we obtain the matrices @xmath113 for the * model ad2 * and @xmath114 for the * model ad3 * for @xmath115 and @xmath116 respectively . the results for these models are presented in figs . [ fig6 ] , [ fig7 ] , [ fig9 ] . by adding shortcut links between pairs of nodes randomly distributed in the upper triangle of matrix @xmath49 we obtain * models ad2s and ad3s * respectively from models ad2 and ad3 . the number of shortcut elements in @xmath49 is taken to be @xmath117 , their amplitude is taken as @xmath118 , after adding shortcuts the columns with shortcut elements are renormalized to unity . thus the sum of elements in each column is equal to unity and @xmath49 belongs to the google matrix class . we note the matrices of these models as @xmath119 respectively for @xmath108 . the results for these models are presented in figs . [ fig8 ] , [ fig9 ] , [ fig10 ] . by replacing matrix elements in the model ad2 by blocks @xmath29 of size @xmath31 ( see sec.2.1 ) we obtain the * model ad2z. in a similar way for the model ad2s we obtain the model ad2zs * with block shortcuts . in this case we restrict our studies only for dimension @xmath115 since the matrix size becomes too large for @xmath120 . amplitudes @xmath121 and @xmath54 are defined as for the models ad2 and ad2s . since the transitions are now given by blocks then the parameter @xmath57 is now defined as @xmath122 with @xmath115 . the results for models ad2z and ad2zs are presented in figs . [ fig11 ] , [ fig12 ] . distribution of ipr @xmath123 on @xmath124plane for the anderson type models ad2 at @xmath115 ( a ) and ad3 at @xmath120 ( b ) ( sec.2.4 ) and the anderson type models with shortcuts ad2s ( c ) and ad3s ( d ) at @xmath125 ( sec.2.5 ) . here @xmath126 ; @xmath127 for ( a , c ) ; @xmath128 for ( b , d ) and @xmath129 , @xmath130 for ( c , d ) . color bars show the ratio @xmath131 ( ipr values are averaged inside cells of coarse - grained lattice @xmath132 ) . , title="fig : " ] + we use exact numerical diagonalization for analysis of spectrum and eigenstates of models of sec.2 . the matrix size @xmath6 is changed from a minimal @xmath133 up to maximal @xmath134 . for the description of the decay of pagerank probability we use a fit @xmath135 which gives us the pagerank exponent of algebraic @xmath68 . in all simulations we use @xmath136 . the right eigenstates @xmath137 of @xmath0 are determined by @xmath138 we characterize @xmath137 by the inverse participation ratio ( ipr ) @xmath139 . this quantity is broadly used in the studies of anderson localization @xcite and determines the number of sites effectively populated by an eigenstate . the value of @xmath36 is independent of normalization . we use normalization @xmath140 for the pagerank eigenstate at @xmath141 . for each eigenvector @xmath142 we can order all nodes in a monotonically decreasing order of @xmath143 thus obtaining the local rank index @xmath144 for a given @xmath142 . such a ranking was used in @xcite . below we describe the results for models of sec.2 . for the model rmz3 at @xmath145 the spectrum is shown in fig . we see that it has a form of @xmath146rays star typical for the directed networks studied in @xcite . the size of the star is slightly reduced since all @xmath147 for @xmath148 , except @xmath141 @xcite . there is also addition reduction of @xmath149 due to finite coupling terms @xmath150 but this reduction is moderate and the spectrum of @xmath53 is close to the spectrum of independent @xmath31 blocks found in @xcite . thus rmz3 model captures a part of real properties of directed networks . an interesting property of eigenstates becomes visible from the dependence of @xmath36 on @xmath37 shown in fig . [ fig1]c at @xmath151 . many eigenstates have relatively small @xmath152 which remain bounded with the increase of @xmath6 up to @xmath153 ( data not shown ) . however , there is a group of states ( gray band ) with @xmath154 growing linearly with @xmath6 ( data not shown ) . these are delocalized states . their origin becomes clear from the following consideration . we can use the anzats in which the elements of @xmath155 are constant inside a given block @xmath156 with a values @xmath157 . then eq.([eq4 ] ) takes the form @xmath158 since the matrix @xmath0 is bistochastic with sum of elements in rows being unity since @xmath159 . the spectrum @xmath35 in ( [ eq5 ] ) is real . thus we obtain in ( [ eq5 ] ) the bloch equation with plane wave delocalized solutions well known for crystals @xcite . these solutions belong to the gray band part of the spectrum in fig . another part of the spectrum corresponds to such @xmath155 which have different values on a scale of one block @xmath29 . the case with different @xmath160 ( e.g. @xmath161 in fig . [ fig1]b , d ) we can use the same anzats for the left vector @xmath157 that leads to the eigenvalue equation : @xmath162 such a problem corresponds to the case of off - diagonal disorder in the @xmath163 anderson model where the localization length , and hence ipr , is diverging at the center of the band @xcite . the spectrum @xmath35 in ( [ eq6 ] ) is real . a similar problem is known as the sinai walk @xcite where transition probabilities on a markov chain are fluctuating . this model has been studied extensively ( see e.g. @xcite and refs . therein ) . the spectrum @xmath35 in ( [ eq6 ] ) , corresponding to this anzats , is the same for the right eigenvectors @xcite . the right eigenvectors are different from the left ones but have a similar structure on average . the ipr values , shown in fig . [ fig1]d are significantly reduced , comparing to the case @xmath164 , except those with @xmath35 close to unity . when @xmath6 is increasing we find that ipr is growing only for @xmath165 while for @xmath166 ipr values remains finite . this corresponds to the known results for the anderson model with off - diagonal disorder . other eigenstates for which @xmath167 is not constant inside @xmath29 blocks correspond to the eigenstates with rather small ipr values @xmath168 . even if the spectrum and eigenstates have interesting properties in the two above cases of model rmz3 there is a weak point here : the pagerank probability @xmath169 in these cases is flat being practically independent of @xmath144 and @xmath154 . thus the situation is very different from the real directed networks with @xmath170 ( see e.g. @xcite ) . this happens due to a space homogeneous structure of the matrix @xmath0 ( a part of fluctuations ) and thus there is no leading node with a large number of links . due to that we try to introduce shortcut links as described in the next sec.3.2 . the spectrum and ipr dependence for rmz3s model with shortcuts are shown in fig . [ fig2 ] for two typical values of parameter @xmath57 . we see that at small values of @xmath171 ( e.g. @xmath60 ) the spectrum structure is practically the same as for rmz3 model . however , for larger values ( @xmath61 ) the size of the spectrum star is decreasing . the values of ipr are significantly reduced at finite values of @xmath57 and our data show that the maximal @xmath36 values remain less than @xmath172 even for the largest size @xmath101 for @xmath173 for all @xmath93 ( data not shown ) . thus in this model all eigenstates remain localized . even if all states are localized the decay of pagerank is more close to the case of real directed networks . indeed , the data of fig . [ fig3]a , b show that @xmath63 have approximately algebraic decay with pagerank index . the fit allows to determine the pagerank exponent @xmath68 which is small at @xmath174 and is growing with increase of @xmath57 reaching values @xmath175 at @xmath176 . it is important to note that @xmath68 is independent of @xmath6 at large @xmath6 values . thus the homogeneous random elements in the upper triangle of @xmath49 matrix allow to obtain @xmath68 close to unity at large @xmath57 . indeed , in the limit of rather large @xmath57 we come to the case of triangular matrix @xmath49 studied in @xcite ( and also in @xcite ) where one obtains an approximate decay @xmath81 . indeed , at large @xmath57 a sum of elements in a row of @xmath0 drops approximately as @xmath177 ( where @xmath144 is a row index ) leading to @xmath81 . indeed , we can say that @xmath178 , where @xmath179 is a homogeneous initial vector , considering this as one iteration of the pagerank algorithm @xcite . we note that for the pagerank vector we have @xmath154 for @xmath180 . thus the model rmz3s has a reasonable spectrum structure and an algebraic pagerank probability decay . but all eigenstates with @xmath93 remain localized . thus we go to the analysis of rmz3f model . the spectrum and ipr values for the rmz3f model are presented in fig . we see that the star spectrum structure is preserved but ipr values are increased in a vicinity @xmath181 . the examples of @xmath63 and @xmath182 dependencies are shown in fig . [ fig3]c , d . it is important to note that @xmath68 is independent of @xmath6 at large @xmath6 values . qualitatively , the situation is similar to the model rmz3s but the effect of @xmath57 on localization properties of @xmath36 is more complicated . indeed , it is well seen in fig . [ fig5]a that the maximal ipr values ( excluding pagerank vectors ) are at first reduced with an increase of @xmath57 from @xmath183 up to @xmath184 but they are increased when @xmath57 goes from @xmath184 to @xmath185 . the dependence of maximal @xmath36 on @xmath6 at @xmath186 is shown in fig . [ fig5]b , c for @xmath187 . we fit this dependence by a power law @xmath188 and obtain for @xmath102 : @xmath189 ( at @xmath60 ) and @xmath190 ( at @xmath78 ) ; for @xmath103 : @xmath191 ( at @xmath60 and @xmath116 ) . these results show that there are certain states ( except pagerank ) that become delocalized in the limit of large matrix size . in a certain sense , for the dependence @xmath192 we have a certain similarity with the results obtained in @xcite where a sub - polynomial growth of @xmath36 with @xmath6 has been found for randomized university networks and preferential attachment models . however , for the rz3f model the spectrum has no large gap and is more similar to the real directed networks . the investigations of rmz3f model at larger sizes ( e.g. with the help of the arnoldi method @xcite ) can provide a more firm conclusion about the delocalization properties of eigenstates in this model . the spectra of ad2 and ad3 models are shown in fig . [ fig6]a , b with color plot of ipr values . we see that there are rather large values of @xmath36 indicating existence of delocalized eigenstates . indeed , a more detailed analysis presented in fig . [ fig7 ] shows that for the states of the spectral range @xmath193 iprs are growing with @xmath6 clearly demonstrating delocalization . indeed , for maximal @xmath36 from this range ( excluding pagerank ) we find @xmath194 at @xmath115 and @xmath195 at @xmath120 . at the same time in a vicinity of the spectrum edge @xmath196 we have @xmath197 for @xmath198 clearly showing that in this part of the spectrum the eigenstates are well localized . indeed , for these localized states we have an exponential decay @xmath199 with the eigenstate rank index @xmath144 ( see fig . [ fig9]a , b ) . such a decay also appears for the localized states of the anderson model in dimension @xmath106 . but for the majority of eigenstates we have significant growth of @xmath36 with @xmath6 showing that these states are delocalized . of course , the case of @xmath115 should be studied in more detail since for the standard anderson model at @xmath115 ( [ eq3 ] ) all eigenstates are exponentially localized @xcite . however , we have here non - hermitian matrix and for our knowledge there are no rigorous results about localization in such matrices in @xmath115 . even if in ad2 , ad3 models we find delocalization , the pagerank in these models is practically flat due to absence of central node ( see fig . [ fig9]a , b ) . another weak point of ad2 , ad3 models is a relatively narrow distribution of eigenvalues with @xmath200 and due to that we continue our analysis with the next model . dependence of @xmath36 on @xmath37 ( a , c ) and @xmath36 on @xmath6 ( b , d ) for the models ad2 ( a , b ) and ad3 ( b , d ) ( see sec.2.4 ) . for ad2 : panel ( a ) is for @xmath133 ( blue , @xmath201 realizations ) and @xmath202 ( red , @xmath203 ) ; panel ( b ) shows dependence @xmath192 with fits @xmath188 for eigenstates at the spectrum edge with @xmath204 ( triangles , @xmath205 ) and for maximal @xmath36 ( circles , @xmath206 ) . for ad3 : panel ( c ) is for @xmath207 ( blue , @xmath201 realizations ) and @xmath208 ( red , @xmath203 ) ; panel ( d ) shows dependence @xmath192 for eigenstates at the spectrum edge with @xmath204 ( triangles , @xmath209 ) for maximal @xmath36 ( circles , @xmath210 ) . here the fits are shown by gray lines , green ( b , d ) and blue , red ( c , d ) dashed lines show dependence @xmath41 . for panels ( b , d ) the number of realizations changes from @xmath201 to @xmath116 when @xmath6 changes from minimal to maximal value . ] the spectra of ad2s , ad3s models are shown in fig . [ fig6]c , d . we see that the additional terms in upper triangle of matrix @xmath49 produce a broadening of @xmath211 which however still remains relatively narrow ( @xmath212 ) . the ipr values are growing with @xmath6 except of the eigenstates at the spectral edge @xmath213 ( see fig . [ fig8 ] ) . for these localized states the exponent @xmath214 is practically zero while for the maximal ipr ( except pagerank ) we find rather large values of @xmath215 at @xmath115 , @xmath216 at @xmath120 . thus , in these models we clearly have the anderson type transition from localized to delocalized eigenstates . in analogy with the 3d anderson model @xcite , we make a conjecture that in models ad2 , ad3 , ad2s , ad3s there is a certain mobility edge curve in the complex plane @xmath35 which separates localized from delocalized states . in a qualitative manner such a curve is visible in fig . [ fig6 ] as a border between blue color of localized states with small @xmath36 and white color of states with large @xmath36 . but definitely more detailed studies are required for a more exact determination of such a mobility edge curve . same as in fig . [ fig7 ] but for the models ad2s ( a , b ) and ad3s ( c , d ) ( see sec.2.5 ) at @xmath125 ; all parameters are as in fig . the fits give : ( b ) @xmath217 at the spectrum edge around @xmath213 ( triangles ) , @xmath218 for maximal @xmath36 ( circles ) ; ( d ) @xmath219 at the spectrum edge around @xmath213 ( triangles ) , @xmath220 for maximal @xmath36 ( circles ) . here @xmath221 , @xmath130 . for panel ( b ) [ ( d ) ] the number of realizations changes from @xmath201 to @xmath116 [ @xmath2 ] when @xmath6 changes from @xmath223 to 16900 [ @xmath224 . , title="fig : " ] 0.4 cm examples of pagerank probability decay are shown in fig . the new element , appearing in ad2s , ad3s models ( comparing to ad2 , ad3 cases ) , is a dependence of the pagerank exponent @xmath68 on the parameter @xmath57 as shown in fig . [ fig10 ] . these data demonstrate that @xmath225 increases from @xmath226 at @xmath227 up to @xmath23 at @xmath92 . thus ad2s , ad3s models have delocalized eigenstates and the pagerank exponent of real directed networks . the only weak point is a narrow distribution of spectrum in @xmath211 . to improve this feature we study in next section the models ad2z , ad2zs . dependence of eigenvector amplitudes @xmath228 on their rank index @xmath144 for models ad2 ( a ) , ad3 ( b ) from ( sec.2.4 ) and ad2s ( c ) , ad3s ( d ) from ( sec.2.5 ) . here @xmath229 for ( a , b ) and @xmath125 for ( c , d ) ; @xmath230 for ( a , c ) and @xmath231 for ( b , d ) . we use @xmath126 and @xmath59 in ( c , d ) . data show maximally delocalized ( maximal @xmath36 corresponding to pagerank , magenta upper curve ) and maximally localized ( smallest @xmath36 , blue bottom curve ) eigenstates . , title="fig : " ] + 0.4 cm dependence of the pagerank exponent @xmath68 on the parameter @xmath57 for the models ad2s ( left panel ) and ad3s ( right panel ) . left panel : the solid curve shows data for @xmath232 , with triangles for @xmath233 and circles for @xmath234 . right panel : the solid curve shows data for @xmath231 , with triangles for @xmath235 and circles for @xmath236 . here @xmath237 and @xmath59 , title="fig : " ] + the spectra of as2z , ad2zs models are shown in fig . [ fig11 ] . we see that the star structure appears due to introduction of blocks @xmath31 . the dependence of ipr @xmath36 on @xmath37 clearly shows the existence of two groups of states with small @xmath238 , presumably for localized phase , and large @xmath239 , presumably for delocalized phase . spectrum @xmath35 ( a , b ) and ipr @xmath36 vs. @xmath37 ( c , d ) for the models ad2z ( a , c ) and ad2zs at @xmath240 ( b.d ) from sec.2.6 . here @xmath241 , @xmath126 and @xmath59 in ( c , d ) . , title="fig:"]0.4 cm the distribution of @xmath36 on @xmath124plane is shown in fig . [ fig12]a , b . again we see signs of the mobility edge curve separating localized ( blue ) and delocalized ( white ) eigenstates . the dependence of @xmath36 on @xmath6 is shown in fig . [ fig12]c . there are well localized states with @xmath36 practically independent of @xmath6 ( @xmath242 ) and delocalized states for which @xmath36 is growing with @xmath6 with a relatively large growth exponent @xmath243 at @xmath229 and @xmath244 at @xmath240 . this gives a strong argument for existence of the anderson transition with a mobility edge in a complex @xmath124plane in these models . the decay of pagerank probability is shown in fig . [ fig12]d : at @xmath229 we have a flat @xmath63 distribution with the exponent @xmath245 , while at @xmath240 we find @xmath246 being close to the values found in real directed networks ( e.g. for the twitter network @xmath247 @xcite ) . thus we can say that the model ad2zs is the one being most close to real directed networks with the number of interesting features : algebraic decay of pagerank probability with the exponent @xmath248 , absence of spectral gap at @xmath249 , a broad star like distribution of eigenvalues in the complex @xmath124plane , existence of localized and delocalized eigenstates of the google matrix with strong indications on the anderson transition and the mobility curve in @xmath124plane . top panels show distribution of ipr @xmath36 values on @xmath124plane for models ad2z ( a ) and ad2zs at @xmath240 ( b ) of sec.2.6 with parameters of fig . [ fig11 ] ; color bar gives the ratio @xmath250 obtained from cells as in fig . panel ( c ) : dependence of @xmath36 on @xmath6 for ad2z with triangles for states with @xmath35 located in the delocalized domain @xmath251 ( red triangles , fit gives @xmath243 ) and in the localized domain @xmath252 ( blue triangles , @xmath253 ) ; for ad2zs at @xmath240 with circles for states with @xmath35 located in the delocalized domain @xmath254 ( red circles , @xmath244 ) and in the quasi - localized domain @xmath252 ( blue circles , @xmath255 ) ; fits are shown by lines , green line shows @xmath41 . panel ( d ) : dependence of pagerank probability @xmath169 on pagerank index @xmath144 for models ad2z ( gray symbols ) and ad2zs at @xmath240 ( black symbols ) ; the fits for the range @xmath256 are shown by dashed lines with @xmath257 ( ad2z ) and @xmath258 ( ad2zs ) for the parameters of panels ( a , b ) . , title="fig : " ] + -0.4 cm we expect that a similar model ad3zs constructed in dimension @xmath120 from the ad3s model will have even stronger delocalization properties . in this work we described various random matrix models of the google matrix of directed networks . our results show that for certain models ( like ad2zs ) we have an algebraic decay of pagerank probability with the exponent @xmath259 , absence of spectral gap at @xmath249 , existence of the anderson transition and mobility edge in the complex @xmath124 plane . we think that the further analysis of the models described here will allow to establish more close links between the anderson delocalization phenomenon in disordered solids and delocalization of eigenstates of the google matrix of directed networks . we thank l.ermann and k.m.frahm for useful discussions . this research is supported in part by the ec fet open project `` new tools and algorithms for directed network analysis '' ( nadine @xmath260 288956 ) . the research of ovz was partially supported by the ministry of education and science of russian federation . anderson , phys . rev . * 109 * , 1492 ( 1958 ) . e. akkermans , and g. montambaux , mesoscopic physics of electrons and photons ( cambridge university press , cambridge uk , 2007 ) . f. evers , and a.d . mirlin , rev . phys . * 80 * , 1355 ( 2008 ) .	we introduce a number of random matrix models describing the google matrix @xmath0 of directed networks . the properties of their spectra and eigenstates are analyzed by numerical matrix diagonalization . we show that for certain models it is possible to have an algebraic decay of pagerank vector with the exponent similar to real directed networks . at the same time the spectrum has no spectral gap and a broad distribution of eigenvalues in the complex plain . the eigenstates of @xmath0 are characterized by the anderson transition from localized to delocalized states and a mobility edge curve in the complex plane of eigenvalues .
it has been shown that well - behaved spacetimes are able to induce an exponential enhancement of the vacuum fluctuations of some nonminimally coupled free scalar fields @xcite . this `` vacuum awakening effect '' may be seen as the quantum counterpart of the classical instability experienced by these fields under linear perturbations @xcite . the exponential growth of the vacuum energy density of an unstable scalar field in the spacetime of , e.g. , a neutron star @xcite would necessarily induce the system to evolve into a new equilibrium configuration culminating in the emission of a burst of free scalar particles @xcite ( see also refs . @xcite for related classical analyses ) . conversely , the determination of the mass - radius ratio of observed neutron stars may be used to rule out the existence of whole classes of nonminimally coupled scalar fields @xcite . this has motivated us to investigate how the vacuum awakening effect is impacted by relaxing some symmetries assumed in ref . @xcite . in order to avoid complications in modeling the fluid , the analyses of deviations from sphericity @xcite and staticity @xcite were considered in the context of massive thin shells . in this paper , we investigate the vacuum awakening mechanism when we endow a spherical shell with electrical charge . the presence of charge affects both the energy conditions satisfied by the shell matter and the effective potential appearing in the radial part of the scalar field equation . hence , it is interesting to inquire how previous results for massive thin shells are modified in the charged case . the paper is organized as follows . in sec . [ sec : spacetime ] , we introduce the spacetime of a massive charged shell . in sec . [ sec : effect ] , we quantize the real scalar field in this background and discuss the vacuum awakening effect . in sec . [ sec : vacuum ] , we investigate the exponential growth of the vacuum energy density . section [ sec : finalremarks ] is dedicated to our final remarks . we assume metric signature @xmath0 and natural units in which @xmath1 unless stated otherwise . let us write the line element describing the spacetime of a spherically symmetric thin shell with mass @xmath2 and electric charge @xmath3 lying at the radial coordinate @xmath4 as @xcite @xmath5 and @xmath6 with @xmath7 where @xmath8 when @xmath9 ( while @xmath10 may assume any positive value when @xmath11 ) and @xmath12 labels quantities defined at @xmath13 , respectively . the three - dimensional timelike surface @xmath14 defined at @xmath4 will be covered with coordinates @xmath15 . the metrics @xmath16 and @xmath17 on @xmath14 as induced from the internal- and external - to - the - shell spacetime portions , respectively , satisfy @xmath18 as demanded by the continuity condition . the shell stress - energy - momentum tensor @xmath19 can be computed from the discontinuity of the extrinsic curvature @xmath20 across @xmath14 ( see , e.g. , ref . @xcite ) . in eq . ( [ tmunu ] ) , @xmath21 where @xmath22 , @xmath23 are the components of the coordinate vectors @xmath24 defined on @xmath14 , and @xmath25 ( inside the delta distribution ) is the proper distance along geodesics intercepting orthogonally @xmath14 ( with @xmath26 , @xmath27 , and @xmath28 inside , on , and outside @xmath14 , respectively ) . by using eqs . ( [ ds_in ] ) and ( [ ds_out ] ) , we obtain @xmath29 and @xmath30 . \label{s22}\ ] ] in order to unveil the most realistic shell configurations , we have investigated the behavior of the different energy conditions with respect to the shell parameters @xmath31 and @xmath32 . ( for the stability of charged shells under linear perturbations , see ref . @xcite . ) we can see that the weak , strong , and dominant energy conditions are simultaneously satisfied in a large portion of fig . [ fig1:ec ] , namely , region 3 . and @xmath32 out of the black area . the black area is excluded because there are no static shell configurations in this region . the weak energy condition is satisfied everywhere but in region 4 . the strong energy condition is violated in regions 2 and 4 , while the dominant energy condition is violated in regions 1 and 4 . we see , thus , that in a large portion , namely , region 3 , the three energy conditions are satisfied . ] now , we consider a nonminimally coupled massless real scalar field @xmath33 satisfying the klein - gordon equation @xmath34 in the spacetime of our spherically symmetric charged shell , where @xmath35 and @xmath36 is the scalar curvature . we follow the canonical procedure and expand the corresponding field operator @xcite @xmath37 , \label{phi_expandido}\ ] ] in terms of positive , @xmath38 , and negative , @xmath39 , norm modes with respect to the klein - gordon inner product , where @xmath40 is a measure defined on the set of quantum numbers @xmath41 . the annihilation @xmath42 and creation @xmath43 operators satisfy @xmath44 = \delta ( \alpha , \beta)$ ] and @xmath45 = 0 $ ] , where the delta function is defined by @xmath46 for @xmath47 , and the vacuum state @xmath48 must satisfy @xmath49 for all @xmath41 . the spacetime symmetries drive us to look for positive - norm modes in the form @xmath50 where @xmath51 represent the spherical harmonic functions , @xmath52 , @xmath53 , @xmath54 , and @xmath55 ( positive - norm condition ) . by using eq . ( [ mode_form ] ) in eq . ( [ kg_eq ] ) , we obtain that @xmath56 and @xmath57 satisfy @xmath58 @xmath59 and @xmath60 depending on the sign of @xmath61 , eq . ( [ t_eq ] ) supplemented by condition ( [ positive - norm condition ] ) admits two kinds of general solutions : @xmath62 for @xmath63 and @xmath64 for @xmath65 . the combination chosen on the right - hand side of eq . ( [ t_exp_form ] ) guarantees that the corresponding @xmath66 functions will be indeed positive - norm modes @xcite . the modes ( [ mode_form ] ) with @xmath57 as given by eqs . ( [ t_osc_form ] ) and ( [ t_exp_form ] ) will be denoted as @xmath67 and @xmath68 corresponding to the usual time - oscillating and tachyonic " modes , respectively . the field - operator expansion ( [ phi_expandido ] ) can be cast , then , in terms of oscillatory and tachyonic modes as @xmath69 \nonumber \\ & + \sum_{l\mu \omega } [ \hat c_{\omega l \mu } w_{\omega l \mu } + \hat c_{\omega l \mu}^\dagger w^_{\omega l \mu } ] , \label{phi_expandido_spherical}\end{aligned}\ ] ] where the only nonzero commutation relations between the creation and annihilation operators are @xmath70 & = \delta_{ll ' } \delta_{\mu \mu'}\delta ( \omega-\omega ' ) , \\ [ \hat c_{\omega l \mu } , \hat c^\dagger_{\omega ' l ' \mu ' } ] & = \delta_{ll ' } \delta_{\mu \mu'}\delta_{\omega \omega'}.\end{aligned}\ ] ] the presence of tachyonic modes makes the scalar field unstable . quantum fluctuations and , consequently , the expectation value of the stress - energy - momentum tensor , grow exponentially in time in the presence of these modes ( see also refs . @xcite for further discussions on the quantization of unstable linear fields in static globally hyperbolic spacetimes ) . now , we shall discuss in detail the radial part of modes @xmath71 and @xmath72 , namely , @xmath73 and @xmath74 , respectively . for the sake of convenience , we define the coordinates @xmath75 such that for @xmath76 @xmath77 while for @xmath78 @xmath79 @xmath80 and @xmath81 \nonumber \\ & + & m\ln[r^2 f(r ) ] + d^{(3 ) } , \;\ ; { \rm for}\;\ ; \|q\| > m , \label { chi+1}\end{aligned}\ ] ] where @xmath82 , @xmath83 , and @xmath84 and @xmath85 are constants chosen such that @xmath86 and @xmath87 fit each other continuously on the shell . in terms of the coordinates @xmath88 , eqs . ( [ f-_eq ] ) and ( [ f+_eq ] ) can be rewritten as schrdinger - like " equations for @xmath89 where @xmath90 : @xmath91 and @xmath92 with @xmath93 and @xmath94 we note that the discontinuity of the potential across the shell is @xmath95 and that solutions @xmath96 must vanish asymptotically to be normalizable : @xmath97 now , let us discuss the junction conditions on @xmath89 for @xmath90 . by using einstein equations in conjunction with eq . ( [ tmunu ] ) , we have that @xmath98 where , by using eqs . ( [ s00 ] ) and ( [ s22 ] ) , we have @xmath99.\ ] ] by using eq . ( [ r ] ) in eq . ( [ kg_eq ] ) , we obtain the junction conditions for @xmath100 and for its first derivative along the direction orthogonal to @xmath14 , namely , @xmath101_{\cal{s } } = 0,\ ] ] and @xmath102_{\cal{s } } = \xi \gamma ( \psi_{\lambda l}/r)_{\cal{s}},\ ] ] where @xmath103 and @xmath104 because of eq . ( [ continuity ] ) . next , we shall look for the conditions on @xmath105 , @xmath32 , and @xmath106 which allow for the existence of solutions @xmath107 for eq . ( [ schrodinger_type_eq2 ] ) associated with regular modes @xmath108 [ see eq . ( [ mode_form - spherical2 ] ) ] . the regularity requirement demands that @xmath109 vanishes at the origin . the fact that the klein - gordon inner product fixes the mode normalization up to a multiplicative phase , allows us to assume without loss of generality that @xmath109 approaches zero from positive values : @xmath110 then , by using eqs . ( [ schrodinger_type_eq2 ] ) and ( [ psi_at_origin ] ) together with the junction condition ( [ continuity ] ) and the fact that @xmath111 [ see eq . ( [ v- ] ) ] , we conclude that @xmath112 while we see from eq . ( [ discontinuity ] ) that in general @xmath113 . from the usual analysis of bound states in nonrelativistic quantum mechanics , it is straightforward to infer that for a fixed @xmath114 the @xmath115 solution of eq . ( [ schrodinger_type_eq2 ] ) ( if any ) satisfying conditions ( [ psi_at_infinity ] ) and ( [ continuity])([psi_at_origin ] ) with the @xmath116_th _ largest @xmath117 will possess @xmath118 zeros for @xmath119 . in particular , for those configurations admitting one single @xmath115 solution , we will have @xmath120 ( with the equality holding at the origin ) . the boundary separating configurations possessing and not possessing tachyonic modes for a fixed @xmath114 will be given by the set @xmath105 , @xmath32 , and @xmath106 , which allows for a single well - behaved solution @xmath115 with @xmath121 . these _ marginal solutions _ @xmath122 will also satisfy conditions ( [ continuity])([psi_at_origin ] ) . the asymptotic behavior of marginal solutions can be obtained directly from eq . ( [ f+_eq ] ) : @xmath123 whose general solution can be cast as @xmath124 with @xmath125 being constants . we see that , for @xmath126 , condition ( [ psi_at_infinity ] ) is actually not satisfied by any nontrivial @xmath127 . this is an artifact which appears because marginal solutions have @xmath121 . the presence of any small @xmath128 would eventually dominate for large enough @xmath129 changing the asymptotic form of eq . ( [ aux2 ] ) to @xmath130 in which case the general solution could be written as @xmath131 where for small enough @xmath117 @xmath132 @xmath133 the fact that eq . ( [ psi_at_infinity ] ) demands @xmath134 implies that we should look for marginal solutions ( [ aux3 ] ) behaving asymptotically as @xmath135 . it is easy to see that @xmath136 is a marginal solution in the _ interior _ of the shell satisfying condition ( [ psi_at_origin ] ) for any charge value . now , we note from eqs . ( [ v-])-([v+ ] ) that the smaller the @xmath114 the deeper " the effective potential @xmath137 making tachyonic modes with vanishing angular momentum the most likely ones to exist . because the appearance of a single tachyonic mode is enough to render the system unstable , we shall restrict our quest to marginal solutions with @xmath126 . then , from eq . ( [ solution_in ] ) , we immediately write @xmath138 for all @xmath3 . next , we look for marginal solutions _ external _ to the shell . for the sake of clarity , we present the cases @xmath139 , @xmath140 , and @xmath141 , separately . case @xmath139 : * the general marginal solution external to the shell for arbitrary @xmath114 can be written as @xcite @xmath142 where @xmath143 guarantees that @xmath144 is a real function , @xmath145 and @xmath146 are legendre functions of the first and second kinds @xcite , respectively , and we note that @xmath147>0 $ ] everywhere outside the shell . then , for @xmath126 @xmath148 now , we impose the constraints ( [ continuity ] ) and ( [ discontinuity ] ) on the solutions ( [ solution_in , l=0 ] ) and ( [ solution_outq < m , l=0 ] ) obtaining the following relationships among the integration constants : @xmath149 and @xmath150}. \label{c10/a10}\ ] ] next , we recall the discussion below eq . ( [ aux4 ] ) and impose @xmath151 to render the asymptotic form of the marginal solution as desired . this establishes a relationship between the coupling constant @xmath106 and the shell parameters @xmath105 , @xmath32 , which allows us to draw the boundary of the unstable regions in figs . [ fig2:qlessm_0 ] and [ fig3:qlessm_0_5 ] for @xmath152 and @xmath153 , respectively . * case @xmath140 : * let us write the general marginal solution external to the shell in this case as @xmath154 with @xmath155 being constants . in the particular case where @xmath126 , we obtain @xmath156 now , by imposing the constraints ( [ continuity ] ) and ( [ discontinuity ] ) on eqs . ( [ solution_in , l=0 ] ) and ( [ solution_outq = m , l=0 ] ) , we obtain @xmath157 the boundary of the unstable region in fig . [ fig4:qeqm ] is obtained by demanding @xmath158 . * case @xmath141 : * finally , the general marginal solution external to the shell in this case can be written as @xmath159 where @xmath160 are constants and @xmath144 are real functions . we then make use of eq . ( 8.834 - 2 ) of ref . @xcite to show that the legendre functions of the second kind can be cast as @xmath161 where @xmath162 now , by choosing the branching cut for @xmath163 along the line @xmath164 , @xmath165 , we write @xmath166 from which it can be seen that @xmath167 as before , we take @xmath126 in eq . ( [ solution_outq > m ] ) , obtaining @xmath168 \\ & \stackrel{r \gg q}{\sim } & b^{(3)}_0 r + c^{(3)}_0 \sqrt{q^2-m^2}. \nonumber\end{aligned}\ ] ] again , by imposing constraints ( [ continuity ] ) and ( [ discontinuity ] ) on eqs . ( [ solution_in , l=0 ] ) and ( [ solution_outq > m , l=0 ] ) , we obtain @xmath169\end{aligned}\ ] ] and @xmath170}. \label{c30/a30}\ ] ] the boundary of the unstable regions in fig . [ fig5:mlessq_1_5 ] was drawn by demanding @xmath171 . in figs . [ fig2:qlessm_0][fig5:mlessq_1_5 ] we show the parameter - space region where tachyonic modes exist for @xmath172 , rendering the scalar vacuum unstable . the black areas represent regions where there are no static shell configurations . for any fixed @xmath173 , there is a small enough @xmath174 such that for @xmath175 the boundary of the unstable regions is determined by @xmath176 being independent of the value of @xmath173 . as @xmath105 increases , the influence of the charge on the instability regions becomes more noticeable . in particular , the minimum value of @xmath105 required for some scalar field with @xmath177 to become unstable grows from @xmath178 when @xmath152 to @xmath179 when @xmath180 . scalar fields with @xmath181 are stable in the spacetime of shells with @xmath182 . we recall that for @xmath11 there are static shell configurations for any @xmath105 at the cost of having the weak , strong and dominant energy conditions being violated by shells with large enough @xmath105 . finally , it is interesting to note that although there are shell configurations which allow for the existence of tachyonic modes for the conformal scalar field case , @xmath183 , at least some energy condition will be violated by the corresponding shell . when @xmath184 . the translucent gray area contains static shell configurations with @xmath185 which violate the dominant energy condition . the dark gray areas depict the regions where the vacuum awakening effect " is triggered . the vertical dotted line corresponds to the conformal coupling , @xmath183 . it intercepts the instability island inside the region where the dominant energy condition is violated . this graph should be used as a benchmark for the sake of comparison with figs . [ fig3:qlessm_0_5 ] , [ fig4:qeqm ] , and [ fig5:mlessq_1_5 ] where @xmath186 . ] . the same color convention as in fig . [ fig2:qlessm_0 ] is assumed here . ] . the same color convention as in fig . [ fig2:qlessm_0 ] is assumed here . we see that the instability islands which appear in figs . [ fig2:qlessm_0 ] and [ fig3:qlessm_0_5 ] for @xmath177 are not present in this case . ] . the dark region is not present here in contrast to the previous three figures because for @xmath11 there are static shell configurations for any @xmath105 . on the other hand , the energy conditions are typically more restrictive . the light translucent gray area corresponds to configurations where the strong energy condition is violated . the dark translucent gray region is excluded by the weak , strong , and dominant energy conditions . ] here , we analyze the exponential growth of the vacuum energy density ( as measured by static observers ) @xmath187 to be the no - particle state as defined according to the oscillating modes at the infinite past , where the initial shell parameters , @xmath188 , @xmath189 , and @xmath190 , do not allow for the existence of tachyonic modes . then , by assuming that the shell evolves to the static configuration , characterized by @xmath2 , @xmath3 , and @xmath10 , which allows for the existence of tachyonic modes , we can use the general expression obtained in ref . @xcite to write the leading contribution to the vacuum energy density ( see ref . @xcite for a more comprehensive discussion ) : @xmath191 where @xmath192 is the heaviside step function ( with @xmath25 being the proper distance of the geodesics which orthogonally intercept @xmath14 ) and @xmath193,\ ] ] @xmath194,\end{aligned}\ ] ] and @xmath195 \left(\frac{\psi^+_{\bar{\omega } 0}(\mathsf{r } ) } { \mathsf{r}}\right)^2 \delta(\ell),\end{aligned}\ ] ] are the leading vacuum contributions to the energy density inside , outside , and on the shell , respectively . here , @xmath196 is a positive constant of order one ( see ref . @xcite for more details ) and @xmath197 represents the largest @xmath117 among all tachyonic solutions . we note that although the absolute value of the vacuum energy density increases exponentially in time , the growth is positive at some places and negative at some other ones in such a way that the total vacuum energy is conserved @xcite . eventually , the spacetime must react to the vacuum energy growth leading the whole system to evolve into some final stable configuration with no tachyonic modes . the analysis of the corresponding evolution in the context of semiclassical gravity is well known to be difficult . however , it was recently shown that the scalar field should lose coherence fast enough to allow backreaction to be treated in the much easier context of general relativity @xcite . we have analyzed the stability of a nonminimally coupled free scalar field in the spacetime of charged spherical shells . the impact of the charge on the instability is enhanced for more compact configurations . also , the cases @xmath184 and @xmath11 differ because in the latter case there are static shell configurations for every radius @xmath10 . notwithstanding , some of them may violate the weak , strong and dominant energy conditions . the presence of charge does not alter the fact that spherically symmetric shells which are able to awake the vacuum for conformally coupled scalar fields , @xmath183 , do not satisfy at least the dominant energy condition . finally , we have calculated the expectation value of the vacuum energy density in order to make explicit its exponential growth in time . a. l. , d. v. , and j. s. were partially ( a. l. , d. v. ) and fully ( j. s. ) supported by so paulo research foundation ( fapesp ) under grants no . 2014/26307 - 8 , no . 2013/12165 - 4 , and no . 2013/07105 - 2 respectively , while w. l. , r. m. , and g. m.were fully ( w. l. , r. m. ) and partially ( g. m. ) supported by conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) .	we investigate the stability of a free scalar field nonminimally coupled to gravity under linear perturbations in the spacetime of a charged spherical shell . our analysis is performed in the context of quantum field theory in curved spacetimes . this paper completes previous analyses which considered the exponential enhancement of vacuum fluctuations in the spacetime of massive shells .
the phase diagram of strongly interacting hadronic / partonic matter has been a subject of primary interest in the physics community for decades . at vanishing ( or low ) chemical potentials lattice qcd ( lqcd ) calculations have provided reliable results on the equation of state @xcite and given a glance at the transport properties in particular in the partonic phase . on the other hand high energy heavy - ion reactions are studied experimentally and theoretically to obtain information about the properties of nuclear matter under the extreme conditions of high baryon density and/or temperature . ultra - relativistic heavy - ion collisions at the relativistic heavy - ion collider ( rhic ) and the large hadron collider ( lhc ) at cern have produced a new state of matter , the strongly interacting quark - gluon plasma ( sqgp ) , for a couple of fm / c in volumes up to a few @xmath13 @xmath14 in central reactions . the produced qgp shows features of a strongly - interacting fluid unlike a weakly - interacting parton gas @xcite as had been expected from perturbative qcd ( pqcd ) . large values of the observed azimuthal asymmetry of charged particles in momentum space @xcite , i.e. the elliptic flow @xmath15 could quantitatively be well described by ideal hydrodynamics up to transverse momenta of the order of 1.5 gev / c @xcite . recent studies of qcd matter in equilibrium using lattice qcd calculations @xcite or partonic transport models in a finite box with periodic boundary conditions @xcite have demonstrated that the ratio of the shear viscosity to entropy density @xmath16 should have a minimum close to the critical temperature @xmath17 , similar to atomic and molecular systems @xcite . on the other hand , the ratio of the bulk viscosity to the entropy density @xmath18 should have a maximum close to @xmath17 @xcite or might even diverge at @xmath17 @xcite . indeed , the minimum of @xmath16 at @xmath19 160 mev is close to the lower bound of a perfect fluid with @xmath20 @xcite for infinitely coupled supersymmetric yang - mills gauge theory ( based on the ads / cft duality conjecture ) . this suggests the ` hot qcd matter ' to be the ` most perfect fluid ' @xcite . on the empirical side , relativistic viscous hydrodynamic calculations ( using the israel - stewart framework ) also require a very small @xmath16 of @xmath21 in order to reproduce the rhic elliptic flow @xmath15 data @xcite ; these phenomenological findings thus are in accord with the theoretical studies for @xmath16 in refs . @xcite . whereas shear and bulk viscosities of hot qcd matter at finite temperature @xmath0 presently are roughly known , the electric conductivity @xmath2 is a further macroscopic quantity of interest @xcite since it controls the electromagnetic emissivity of the plasma . first results from lattice calculations on the electromagnetic correlator have provided results that varied by more than an order of magnitude @xcite . furthermore , the conductivity dependence on the temperature @xmath0 ( at @xmath22 ) is widely unknown , too . the electric conductivity @xmath2 is also important for the creation of electromagnetic fields in ultra - relativistic nucleus - nucleus collisions from partonic degrees - of - freedom , since @xmath2 specifies the imaginary part of the electromagnetic ( retarded ) propagator and leads to an exponential decay of the propagator in time @xmath23 @xcite . apart from the electric conductivity the magnetic response of the qgp ( or strong vacuum ) to external magnetic fields @xmath24 has also been of current interest from the experimental side @xcite as well as from lattice qcd @xcite . strong magnetic fields are created in peripheral relativistic nucleus - nucleus collisions by the charges of the spectator protons during the passage time of the nuclei @xcite and at the top rhic energy of @xmath25 = 200 gev magnetic fields of order @xmath26 can be reached . this has lead to the suggestion of a charge separation effect due to the chiral - magnetic - effect ( cme ) in these reactions @xcite . on the other side lqcd has been focusing on the magnetic catalysis of the chiral @xmath27 condensate at very strong @xmath24 fields at low temperature and the inverse magnetic catalysis of the chiral condensate at temperatures close the critical temperature indicating a decrease of @xmath17 for high magnetic fields . note , however , that these studies involve time - independent magnetic fields that are significantly higher than those achieved in peripheral nucleus - nucleus collisions for very short times where an equilibrated qgp might not have been established . nevertheless , a sufficient knowledge of the electric and magnetic response of the qgp ( in equilibrium ) to external electromagnetic fields is mandatory to explore a possible generation of the chiral - magnetic - effect in predominantly peripheral heavy - ion reactions @xcite and to determine the photon production from the qgp in heavy - ion collisions at different centralities and bombarding energies @xcite . in this work we extend our previous studies on the electric conductivity @xmath28 @xcite for ` infinite parton matter ' also to finite quark chemical potential @xmath1 employing the parton - hadron - string dynamics ( phsd ) transport approach @xcite , which is based on generalized transport equations derived from the off - shell kadanoff - baym equations @xcite for green s functions in phase - space representation ( beyond the quasiparticle approximation ) . this approach describes the full evolution of a relativistic heavy - ion collision from the initial hard scatterings and string formation through the dynamical deconfinement phase transition to the strongly - interacting quark - gluon plasma ( sqgp ) as well as hadronization and the subsequent interactions in the expanding hadronic phase . in the hadronic sector phsd is equivalent to the hadron - string - dynamics ( hsd ) transport approach @xcite a covariant extension of the boltzmann - uehling - uhlenbeck ( buu ) approach @xcite that has been used for the description of @xmath29 and @xmath30 collisions from lower schwerionen - synchrotron ( sis ) to rhic energies in the past . on the other hand , the partonic dynamics in phsd is based on the dynamical quasi - particle model ( dqpm ) @xcite , which describes qcd properties in terms of single - particle green s functions ( in the sense of a two - particle irreducible ( 2 pi ) approach ) and reproduces lattice qcd results including the partonic equation of state in thermodynamic equilibrium . for further details on the phsd off - shell transport approach and hadronization we refer the reader to refs . @xcite . the layout of our study is as follows : in section ii we concentrate on calculating the electric conductivity for ` infinite ' qcd matter also at finite quark chemical potential @xmath1 and provide simple parametrizations for the dependence of @xmath2 on @xmath1 . furthermore , we calculate the frequency dependence @xmath9 for periodic external fields and compare the results with those from the dynamical quasiparticle model ( dqpm ) . in section iii we compute the diamagnetic and paramagnetic contributions to the magnetization @xmath31 of the plasma as a function of temperature @xmath0 and @xmath1 and compare to the experimental situation at rhic . a summary and discussion of results is presented in section iv . we briefly recall the setup of our studies within phsd . the infinite hadronic or qcd matter is simulated within a cubic box with periodic boundary conditions at various values for the energy density ( or temperature ) and the quark chemical potential @xmath1 . the size of the box is fixed to @xmath32 @xmath14 as in the previous investigations @xcite . the initialization is done by populating the box with light ( @xmath33 ) and strange ( @xmath34 ) quarks , antiquarks and gluons slightly out of equilibrium . the system approaches kinetic and chemical equilibrium during its time evolution within phsd . for more details on the simulation of equilibrated partonic systems using phsd in the box at finite temperature @xmath0 and quark chemical potential @xmath1 we refer the reader to ref . @xcite . we recall that phsd is an off - shell transport approach that propagates quasi - particles with broad spectral functions . numerically , the continuous spectral distribution in the mass of a particle ( given by its spectral function ) is probed by a large number of test - particles with ( evolving ) masses @xmath35 . in order to include the effects from an external electric field @xmath36 or magnetic field @xmath24 , the propagation of each charged test - particle @xmath37 is performed with the additional lorentz force in the equation of motion : @xmath38 where @xmath39 denotes the fractional charge of the test - particle ( @xmath40 ) and @xmath41 its energy . we recall that the external electric field will lead to an acceleration of positively and negatively charged particles in opposite directions while the particle scatterings / interactions will damp this acceleration and eventually lead to an equilibrium current ( cf . fig . 1 in ref . the electric current density @xmath42 ( for an external electric field in @xmath43-direction ) is calculated by @xmath44 where @xmath35 is the mass of the test - particle @xmath37 at time @xmath45 . the summation in ( [ e2 ] ) is carried out over @xmath46 ensemble members @xmath47 while @xmath48 denotes the time - dependent number of physical ( @xmath49 ) quarks and antiquarks that varies with time @xmath45 due to the processes @xmath50 in a single member of the ensemble ( run ) . the number of runs @xmath46 is typically taken as a few hundred which gives a current @xmath42 practically independent on the number of ensemble members @xmath46 . we recall that ( without external fields ) each run of the ensemble is a microcanonical simulation of the dynamics as inherent in the phsd transport approach which strictly conserves the total four - momentum as well as all discrete conservation laws ( e.g. net fermion number for each flavor etc . ) a note of caution has to be given , since due to an external field we deal with an open system with increasing energy density ( temperature ) in time . therefore we employ sufficiently small external fields @xmath51 , such that the energy increase during the computation time ( in each run ) stays below 2% and the increase in temperature below 1 mev . we find that for constant electric fields up to @xmath52 = 50 @xmath53 a stable electric current @xmath54 emerges that is @xmath55 ( cf . accordingly , we obtain the conductivity @xmath56 from the ratio of the stationary current density @xmath54 and the electric field strength as @xmath57 the results for the dimensionless ratio ( [ e4 ] ) at @xmath58 ( from ref . @xcite ) are displayed in fig . [ figg1 ] as a function of the scaled temperature @xmath59 in comparison to more recent lattice qcd results from refs . @xcite and suggest a minimum in the ratio @xmath60 close to the critical temperature @xmath17 followed by an approximate linear rise up to 2 @xmath17 ( cf . ref . @xcite ) . the most recent lqcd results @xcite are roughly compatible with the phsd calculations . moreover , the lattice data from ref . @xcite give support for the linearity of @xmath6 with temperature above @xmath17 . ( [ e4 ] ) as a function of the scaled temperature @xmath59 for @xmath58 in comparison to recent lattice qcd results : triangles - quenched qcd results in the continuum limit with wilson - clover fermions and renormalized vector currents @xcite , star - quenched su(2 ) lattice gauge theory @xcite , open circle - qcd with two dynamical flavors of wilson - clover fermions @xcite . the phsd results ( full dots ) are the same as in ref . @xcite . ] we now focus on the explicit dependence of @xmath61 as a function of the chemical potential @xmath1 which is shown in fig . [ pic : condtvsmu ] for a fixed temperature @xmath0=200 mev . as a function of the chemical potential @xmath1 for @xmath0= 200 mev from the phsd calculations ( full dots ) . the error bars indicate the statistical uncertainty for the ratio ( [ e4 ] ) when performing calculations for different external field strength @xmath51 up to 50 mev / fm . ] the numerical result can be fitted with a quadratic correction ( solid line in fig . [ pic : condtvsmu ] ) @xmath62 with @xmath63 for @xmath64 gev . this result comes about as follows : we recall that the electric conductivity of gases , liquids and solid states is described in the relaxation time approach by the drude formula @xmath65 where @xmath66 denotes the density of non - localized charges , @xmath67 is the relaxation time of the charge carriers in the medium and @xmath68 their effective mass . this expression can be directly computed for partonic degrees - of - freedom within the dqpm , which was used to match in phsd the quasiparticles properties to lattice qcd results in equilibrium for the equation - of - state ( eos ) as well as various correlators @xcite . we note that the electromagnetic correlator from lqcd calculations @xcite appears to match rather well the back - to - back dilepton rate from phsd at @xmath69 ( cf . fig . 2 in ref . @xcite ) , which suggests that the results of our calculations for @xmath2 - for vanishing invariant mass - should also be close to the lqcd extrapolations from @xcite . in the dqpm , the relaxation time for quarks / antiquarks is given by @xmath70 , where @xmath71 is the width of the quasiparticle spectral function ( cf . furthermore , the spectral distribution for the mass of the quasiparticle has a finite pole mass @xmath72 that is also fixed in the dqpm , as well as the density of ( @xmath73 ) quarks / antiquarks as a function of temperature and chemical potential ( cf . thus , we obtain for the dimensionless ratio ( [ e4 ] ) the expression @xcite @xmath74 where @xmath75 denotes the total density of quarks and antiquarks and the prefactor @xmath76 reflects the flavor averaged fractional quark charge squared @xmath77 . as found in our previous study @xcite the dqpm results match well with the explicit phsd calculations in the box for @xmath1=0 since phsd in equilibrium is a suitable transport realization of the dqpm @xcite . for @xmath78 mev . the solid ( red ) line displays the fit ( [ d5 ] ) to the dqpm results . ] in the dqpm we have @xmath79 and @xmath80 for @xmath81 100 mev , however , @xmath82 with the same coefficient @xmath83 as in eq . ( [ parabol ] ) . this is demonstrated explicitly in fig . [ pic : densityfinitmu ] where the actual dqpm results for the quark+antiquark density ( full dots ) are compared to the fit ( [ d5 ] ) ( solid line ) . the temperature dependence of the expansion coefficient @xmath83 is found to be @xmath84 such that the ratio @xmath6 can be approximated by @xmath85 in fig . [ pic : korrfaktor ] we display the coefficient @xmath86 in the temperature range 170 mev@xmath87 250 mev giving @xmath88 as a best fit . this completes our study on the stationary electric conductivity @xmath2 . in ( [ expand ] ) as a function of the temperature @xmath0 for @xmath89 100 mev . the solid line shows the average value in the interval 170 mev @xmath90 250 mev . ] we now extent our study to external periodic fields of frequency @xmath91 , @xmath92 in this case the electric current density @xmath42 does not achieve a constant equilibrium value and also oscillates with the frequency @xmath91 . [ pic : jzvsw ] shows the time - dependence of the current @xmath42 from phsd for different frequencies as a function of @xmath93 with their amplitudes normalized to one in comparison to the external electric field @xmath94 ( dotted red line ) . for @xmath95 ( solid black ) and @xmath96 ( dash dotted blue ) normalized to the equilibrium amplitude for temperature @xmath97 mev and @xmath98 mev / fm . the dotted red line shows the time - dependence of the external electric field @xmath94 . ] the current @xmath42 is seen to be shifted in phase compared to the electric field ; the phase shift @xmath99 increases with the frequency @xmath91 up to @xmath100 . the currents in fig . [ pic : jzvsw ] can be well described by @xmath101 we find that the amplitude @xmath102 decreases with the frequency @xmath91 since the current has less time to build up and to follow the external field . this behavior is in line with the complex conductivity @xmath9 for oscillating fields , @xmath103 where @xmath104 is the quasi - particle width of the charged particles ( quarks and antiquarks ) . we have computed the current @xmath42 for @xmath97 mev and @xmath105 mev / fm in the frequency range @xmath106 . fig . [ pic : phasevsw ] shows the phase shift @xmath107 and fig . [ pic : ampvsw ] the amplitude @xmath108 ( full dots ) . the phsd results can be easily followed up within the dqpm results ( shown by the red lines ) which provide again a good description of the microscopic calculations . since the complex conductivity @xmath9 depends only on the width @xmath104 and the stationary conductivity @xmath2 in ( [ sigo ] ) its actual values for different temperatures @xmath0 and finite chemical potential @xmath1 follow directly from our previous results in this section . note that for actual electric fields in peripheral au+au collisions at the top rhic energy we have @xmath109 such that the electric conductivity @xmath9 is suppressed relative to its equilibrium value by more than a factor of 100 . over @xmath110 as a function of the frequency @xmath91 from the phsd calculations ( full dots ) for @xmath97 mev . the red line shows the phase shift as expected from the dqpm using ( [ sigo ] ) . ] as a function of the frequency @xmath91 from the phsd calculations ( full dots ) for @xmath97 mev . the red line shows the expected amplitude within the dqpm . ] in order to explore the magnetic response of the partonic system within phsd we will assume the magnetic field to be sufficiently small such that terms @xmath111 can be neglected ( see below ) . note that this limit does not hold for the strong fields @xmath112 0.1 - 1 gev@xmath113 0.5 - 5 gev / fm ) in actual lattice qcd studies @xcite . using @xmath114 with the pauli matrices @xmath115 , the kinetic momentum @xmath116 and the angular momentum @xmath117 the dirac equation can be rewritten for 2-component quark and antiquark spinors leading to the hamiltonian @xmath118 @xmath119 with @xmath120 . in case of small energies @xmath121 this leads to the well known expression for the non - relativistic pauli equation : @xmath122 the change of the energy of the system in the presence of an external magnetic field @xmath24 is determined by the magnetic moment @xmath123 : @xmath124 which has a contribution from the angular momentum @xmath117 of a particle and from the spin @xmath125 . in the following we will investigate both terms separately since they provide contributions to the magnetic moment of opposite sign . in analogy to sec . ii we are dealing with an open system but the increase in the total energy stays below 1% . the induced angular momentum @xmath117 emerges from the lorentz force ( [ e1 ] ) on a charged particle due to an external field @xmath24 , @xmath126 and induces a magnetic moment opposite to the direction of the @xmath127-field since the charged particle spirals around the magnetic field with frequency @xmath128 , where @xmath129 is the momentum component of the particle perpendicular to the direction of the magnetic field and @xmath130 is the radius of the spiral . we obtain alternatively for the angular momentum @xmath131 where @xmath132 is the projection of the momentum on the direction of the magnetic field @xmath133 . inserting the expression for the radius @xmath130 we get @xmath134 assuming the magnetic field to be oriented in @xmath135-direction and employing the lorentz force @xmath136 we end up with @xmath137 this gives the induced magnetic moment @xmath138 since the lorentz force changes only the direction of @xmath116 and not its magnitude @xmath139 the particle energy @xmath140 is conserved , too . as a consequence the energy contribution in the hamiltonian ( [ eq : hamiltondirac ] ) is independent from the magnetic field strength : @xmath141 we note that the diamagnetic contribution can not be seen in approaches that calculate the magnetization by differentiation of the thermodynamic potential ( e.g. free energy f ) with respect to the magnetic field @xmath142 . in principle , the angular momentum @xmath117 has to be quantized . however , the actual values for @xmath117 ( in units of @xmath143 ) are @xmath144 1 for small field strength since @xmath145 such that quantum corrections are subleading in our case . the quark and antiquark spins provide a paramagnetic contribution since the spin precession around the direction of the magnetic field @xmath24 in thermal equilibrium gives a positive magnetic moment @xmath146 since the energy becomes reduced according to eq . ( [ eq : hamiltondirac ] ) . the spin degree - of - freedom is introduced in phsd in line with the generalized test - particle ansatz @xcite for the wightman function @xmath147 @xmath148 where @xmath149 and @xmath150 stand for space - time and four - momentum coordinates , respectively , while @xmath151 denotes the spin degree - of - freedom . in ( [ eq : testspin ] ) the number of ensemble members ( runs ) is denoted by @xmath46 whereas @xmath152 is the number of partons in the run @xmath153 that describe the physical particles in each microcanonical simulation . the spin degree - of - freedom has to be treated in line with quantum mechanics according to the interaction hamiltonian ( [ eq : hamiltondirac ] ) , i.e. @xmath154 the spin - wavefunction for a spin @xmath155 fermion is taken as a 2-component spinor @xmath156 with @xmath157 denoting the probability for the spin in @xmath43-direction ( parallel to the magnetic field ) while @xmath158 stands for the probability for the anti - parallel orientation . we assume the spin - wavefunction to be normalized . i.e. @xmath159 . the projection on the coordinate axis @xmath160 is provided by @xmath161 the time - evolution of the spin projections according to the hamiltonian ( [ hs ] ) can be worked out in a straight forward way using @xmath162 with @xmath163 and eq . ( [ eq : projection ] ) . the resulting equations of motion lead to a precession of the spin of a quark / antiquark with frequency @xmath164 which changes only if the energy of the particle @xmath140 changes in a collision or in the inelastic reaction @xmath165 . since we are using effective interactions , which should be considered as an approximation to the resummed interactions , we can not determine the actual spin of the parton degrees - of - freedom . for this reason we neglect the spin of the gluons and treat the quark spin statistically . in order to describe an equilibration of the spin degree - of - freedom we introduce a spin flip in 1/3 of the elastic collisions , as motivated by nuclear physics , that favors final spin states parallel to the * b-field . in order to simplify the ( time expensive ) calculations we introduce the constraint ( in equilibrium ) @xmath166 @xmath167 which is introduced explicitly in the transition matrix element squared . accordingly , in ( [ eq : spinequil ] ) @xmath150 denotes the probability for a spin flip and @xmath168 the occupation probability for given spin orientation . in practice the probabilities @xmath150 are taken as @xmath169 and lead to the proper equilibrium distribution when neglecting the @xmath170 channels . when including these channels we find numerically deviations from the equilibrium distribution by up to 10% since the gluon channels reduce the spin orientation in the direction of the @xmath24-field , i.e. induce a diamagnetic effect. the actual values of the probabilities ( 29 ) are not important for the present study . as long as they fulfill ( 28 ) the equilibrium distribution changes only slightly within statistical error bars . the same holds for the number of collisions with spin flips . their values are only important for the timescales of the spin equilibration , which , however , we do not address . the magnetization @xmath31 is defined by the spin density of the system as @xmath171 which in case of small magnetic fields @xmath172 - as in our present study - is proportional to the strength of the @xmath142-field thus defining a magnetic susceptibility @xmath173 by @xmath174 in order to explore the range of external magnetic fields @xmath172 we can handle reliably within the phsd calculations for partonic systems we show in fig . [ pic : mulvsb ] the energy contribution to the magnetic field ( [ e9 ] ) as a function of @xmath172 for a temperature @xmath0=190 mev . in fact , the calculations for the energy shift due to the magnetic field @xmath172 give constant results for @xmath175 50 mev / fm - when discarding the spin degrees - of - freedom - while for stronger fields more significant deviations emerge up to @xmath176 10% for @xmath177 200 mev / fm . accordingly , we will restrict to @xmath178 50 mev / fm ( @xmath179 0.01 gev@xmath180 ) in the following . in the phsd calculation ( full dots ) as a function of the field strength @xmath172 for a temperature of @xmath0 = 190 mev at @xmath1=0 . the solid line reflects a constant for small / moderate field strength . ] the temperature dependence of @xmath181 from the phsd calculation is shown in fig . [ pic : mulvst ] by full dots and can be well fitted in the interval 170 mev @xmath182 250 mev by @xmath183\ ] ] where the temperature @xmath0 is given in units of mev . the diamagnetic contribution to the magnetization from the lorentz force on the quarks and antiquarks then can be readily extracted by dividing @xmath184 by the strength of the magnetic field . in the phsd calculation as a function of the temperature @xmath0 for @xmath58 . the numerical uncertainties are smaller than the size of the dots . the solid line shows the fit ( [ fit5 ] ) . ] as a next step we compute the magnetic susceptibility @xmath173 in the phsd calculations according to eq . ( [ eq12 ] ) for different field strength @xmath172 at @xmath1 = 0 . we have found the necessary energy for a spin flip to be very small in comparison to the total energy of the quarks ( @xmath185 ) and therefore have discarded it in our actual simulations . the results for the susceptibility @xmath173 are displayed in fig . [ pic : susceptvsb ] for @xmath0= 190 mev and ( within numerical accuracy ) show a constant value even up to @xmath172= 200 mev / fm . in this case the numerical accuracy increases with the field strength since the spin - flip probabilities in eq . ( [ eq : spinflipbed ] ) differ more significantly for larger magnetic fields . nevertheless , we have a stable numerical window @xmath172 from 25 - 50 mev / fm where the diamagnetic and parametic contributions to the magnetic moment can be calculated with sufficient accuracy . note that the energy shift due to the paramagnetic contribution is given by @xmath186 and decreases quadratically with the field strength . ( [ eq12 ] ) from phsd as a function of the external magnetic field @xmath172 for a temperature @xmath0=190 mev at vanishing quark chemical potential @xmath1 = 0 . ] the temperature dependence of the magnetic susceptibility @xmath187 from phsd is displayed in fig . [ pic : suszeptvst ] by the full dots and can be fitted as @xmath188 with @xmath0 given in mev ( in the interval 170 mev @xmath182 250 mev ) . from phsd ( full dots ) in comparison to the fit ( [ eq13 ] ) for @xmath1=0 . the numerical uncertainties are indicated by the errorbars . ] the total energy shift due to the both interactions with the magnetic field is given by @xmath189 and decreases with @xmath190 at constant temperature @xmath0 . at a critical field @xmath12 the energy shift changes sign , i.e. for @xmath191 the magnetization changes from diamagnetic to paramagnetic with increasing magnitude of the field @xmath24 . ) as a function of the temperature @xmath0 for @xmath1 = 0 from the phsd calculations ( thick solid black line ) . the dashed extensions are based on the extrapolated fits and not explicitely controlled by phsd calculations . the ( lower ) thin solid blue line results when assuming all quark / antiquark spins to be oriented in @xmath24 direction . the constant lower line displays the maximum value for the magnetic field strength as found in ref . @xcite at the top rhic energy . ] this quantity has a minimum ( within phsd ) close to the critical temperature @xmath192 158 mev ( cf . [ pic : bcvst ] ) with a minimum @xmath193 ( thick solid black line - extrapolated by the dashed line according to the fits performed ) . for comparison we also show the limiting results when assuming all quark / antiquark spins to be oriented in @xmath24 direction . this line is slightly lower because the coupling to the gluons ( in phsd ) reduces the paramagnetic contribution to the magnetization ( diamagnetic gluon effect ) . in the qgp phase the critical field @xmath194 rises with temperature and separates the diamagnetic ( below ) from the paramagnetic response ( above ) of the qgp . note that the maximal field strength in peripheral au + au collisions at the top rhic energy @xmath25 = 200 gev was found to be @xmath195 gev@xmath180 @xcite ( constant solid line ) - during the passage time of the nuclei - which is significantly lower than the critical field in fig . [ pic : bcvst ] . accordingly , the response of the qgp in actual heavy - ion experiments should be diamagnetic . however , for the much higher field strength explored in lattice qcd calculations @xcite for temperatures close to @xmath17 the response should be paramagnetic . as in case of the electric conductivity @xmath56 we can also compute the magnetic response at finite quark chemical potential @xmath1 in phsd . in analogy to fig . 2 we find essentially a quadratic dependence on @xmath1 as demonstrated in fig . [ pic : mulvsmu ] for @xmath196 at @xmath0 = 200 mev . this dependence is also obtained for the magnetic susceptibility @xmath10 ( not shown explicitly ) although with larger numerical errorbars . as a function of @xmath1 from phsd ( full dots ) in comparison to the fit ( [ eq20 ] ) in @xmath1 ( solid line ) for @xmath0= 200 mev . ] again we find the temperature dependence of the coefficient to be @xmath197 such that we get the approximations @xmath198 @xmath199 as an example we show the coefficient @xmath200 in fig . [ pic : lkorrfaktor ] for temperatures from 190 to 250 mev . in this temperature intervall the expansion coefficient may be well approximated by @xmath201 = 0.57 . similar statements ( with less accuracy ) hold for the magnetic susceptibility in ( [ eq20 ] ) which gives @xmath202 = 0.49 . the scaling ( [ eq20 ] ) can be traced back again to the scaling of the quark+antiquark density @xmath203 in the dqpm . in eq . ( [ eq20 ] ) for @xmath204 in case of finite @xmath89 100 mev as a function of temperature @xmath0 . ] in conclusion , we have evaluated the electric conductivity @xmath56 of the quark - gluon plasma as a function of temperature @xmath0 and quark chemical potential @xmath1 by employing the parton - hadron - string dynamics ( phsd ) off - shell transport model in a finite box for the simulation of dynamical partonic systems in equilibrium . the phsd approach in the partonic sector is based on the lattice qcd equation of state of @xcite ; accordingly , it describes the qgp entropy density @xmath205 , the energy density @xmath206 and the pressure @xmath207 from lqcd @xcite very well . studies of the qcd matter within phsd have previously given reasonable results also for the shear and bulk viscosities @xmath208 and @xmath209 versus @xmath0 @xcite and related transport coefficients @xcite . in extension of our previous study in ref . @xcite we have found that the electric conductivity shows a simple scaling with @xmath8 ( cf . eq.(8 ) ) which essentially can be traced back to the variation of the quark+antiquark density @xmath210 of the system in the dynamical quasiparticle model ( dqpm ) . we recall that phsd calculations in a fixed box in equilibrium give practically the same results for the equation of state of qcd matter and transport coefficients as the dqpm @xcite . in analogy to the ratio of shear viscosity to entropy density @xmath16 we find a clear minimum of the dimensionless ratio @xmath6 in the vicinity of the critical temperature @xmath17 with values close to more recent lattice qcd calculations @xcite . this prediction as well as the explicit dependence on @xmath1 should be controlled by related qcd studies on the lattice . the frequency dependence of the complex electric conductivity @xmath9 is found to be well in line with the simple results from the dqpm ( 11 ) , too . furthermore , we have explored the response of the partonic system on a moderate external magnetic field @xmath211 gev / fm and studied separately the diamagnetic response from the lorentz force ( 1 ) and the paramagnetic response due to the interaction of the quark spin with the magnetic field in eq . within the temperature range investigated ( 170 mev @xmath212 250 mev ) the magnetic moment due to the lorentz force rises with temperature @xmath0 but drops with @xmath213 providing a constant energy shift due to a coupling to the magnetic field @xmath214 . the coupling of the spin to the @xmath142-field gives a paramagnetic contribution which can well be characterized by a magnetic susceptibility @xmath10 . its contribution to the energy shift of the system is negative and increases with @xmath190 . according , there is a critical magnetic field @xmath12 for which the response of the system changes from diamagnetic to paramagnetic . the actual values for @xmath12 ( cf . fig . 12 ) demonstrate that the response of the qgp in ultrarelativistic heavy - ion collisions should be diamagnetic since the maximal magnetic fields created in these collisions are smaller than @xmath12 . as in case of the electric conductivity the dependence of the energy shift @xmath196 as well as the magnetic susceptibility @xmath10 show a very similar scaling correction @xmath215 with a coefficient of order 0.5 . we close by noting that in the hadronic phase for temperatures below @xmath17 and @xmath1 = 0 the system dominantly consists of pseudoscalar mesons that have no spin and accordingly a low ( or vanishing ) paramagnetic contribution to the magnetization @xmath31 when restricting to the first order in the magnetic field . on the other hand , the charged hadrons see the lorentz force ( 1 ) and build up a diamagnetic contribution . + the authors acknowledge valuable discussions with e. l. bratkovskaya , v. konchakovski , o. linnyk and r. marty during the course of this study which was supported by the loewe center hic for fair . 99 y. aoki _ et al . _ , b 643 * , 46 ( 2006 ) ; s. borsanyi _ et al . _ , jhep * 1009 * , 073 ( 2010 ) ; jhep * 1011 * , 077 ( 2010 ) ; jhep * 1208 * , 126 ( 2012 ) ; phys . lett . b * 370 * , 99 ( 2014 ) . p. petreczky [ hotqcd collaboration ] , pos lattice * 2012 * , 069 ( 2012 ) ; aip conf . proc . * 1520 * , 103 ( 2013 ) . m. gyulassy and l. d. mclerran , nucl . a * 750 * , 30 ( 2005 ) ; e. v. shuryak , nucl . a * 750 * , 64 ( 2005 ) ; u. w. heinz , aip conf . * 739 * , 163 ( 2004 ) ; a. peshier and w. cassing , phys . * 94 * , 172301 ( 2005 ) . j. adams _ et al . _ ( star collaboration ) , nucl . phys . a * 757 * , 102 ( 2005 ) . k. adcox _ et al . _ ( phenix collaboration ) , nucl . phys . a * 757 * , 184 ( 2005 ) . i. arsene _ et al . _ ( brahms collaboration ) , nucl . phys . a * 757 * , 1 ( 2005 ) . b. b. back _ et al . _ ( phobos collaboration ) , nucl . a * 757 * , 28 ( 2005 ) . k. aamodt _ et al . _ ( alice collaboration ) , phys . . lett . * 105 * , 252302 ( 2010 ) . p. huovinen _ et al . b * 503 * , 58 ( 2001 ) . p. f. kolb , p. huovinen , u. heinz , and h. heiselberg , phys . b * 500 * , 232 ( 2001 ) . d. teaney , j. lauret , and e. v. shuryak , phys . lett . * 86 * , 4783 ( 2001 ) . t. hirano and k. tsuda , phys . c * 66 * , 054905 ( 2002 ) . p. f. kolb and r. rapp , phys . c * 67 * , 044903 ( 2003 ) . p. huovinen , in _ quark - gluon plasma 3 _ , edited by r. c. hwa and x .- wang ( world scientific , singapore , 2004 ) ; p. f. kolb and u. w. heinz , edited by r. c. hwa and x .- wang ( world scientific , singapore , 2004 ) . h. b. meyer , phys . d * 76 * , 101701 ( 2007 ) . s. sakai and a. nakamura , pos * lat2007 * , 221 ( 2007 ) . v. ozvenchuk , o. linnyk , m. i. gorenstein , e. l. bratkovskaya , and w. cassing , phys . c * 87 * , 024901 ( 2013 ) . v. ozvenchuk , o. linnyk , m. i. gorenstein , e. l. bratkovskaya , and w. cassing , phys . c * 87 * , 064903 ( 2013 ) . r. a. lacey and a. taranenko , pos * cfrnc2006 * , 021 ( 2006 ) . d. kharzeev and k. tuchin , jhep * 09 * , 093 ( 2008 ) . f. karsch , d. kharzeev , and k. tuchin , phys . b * 663 * , 217 ( 2008 ) . p. romatschke and d. t. son , phys . d * 80 * , 065021 ( 2009 ) . g. d. moore and o. saremi , jhep * 09 * , 015 ( 2008 ) . c. sasaki and k. redlich , phys . c * 79 * , 055207 ( 2009 ) ; nucl . a * 832 * , 62 ( 2010 ) . g. policastro , d. t. son , a. o. starinets , phys . lett . * 87 * , 081601 ( 2001 ) ; p. k. kovtun , d. t. son , a. o. starinets , phys . rev . lett . * 94 * , 111601 ( 2005 ) . t. hirano and m. gyulassy , nucl . phys . a 769 , 71 ( 2006 ) . b. jacak and p. steinberg , phys . today * 53 * , 39 ( 2010 ) . p. romatschke , u. romatschke , phys . lett . * 99 * , 172301 ( 2007 ) . h. song and u. w. heinz , phys . c * 77 * , 064901 ( 2008 ) . m. luzum and p. romatschke , phys . rev . c * 78 * , 034915 ( 2008 ) . b. schenke , s. jeon , and c. gale , phys . c * 82 * , 014903 ( 2010 ) . s. plumari , a. puglisi , f. scardina , and v. greco , phys . c * 86 * , 054902 ( 2012 ) . s. mattiello and w. cassing , eur . j. c * 70 * , 243 ( 2010 ) . y. hirono , m. hongo and t. hirano , arxiv:1211.1114 . s. i. finazzo and j. noronha , arxiv:1311.6675 . d * 83 * , 034504 ( 2011 ) ; o. kaczmarek _ et al . _ , pos confinement x , 185 ( 2012 ) . g. aarts , c. allton , j. foley , s. hands , and s. kim , phys . rev . * 99 * , 022002 ( 2007 ) . s. gupta , phys . b * 597 * , 57 ( 2004 ) . v. buividovich _ et al . _ , lett . * 105 * , 132001 ( 2010 ) . b. b. brandt , a. francis , h. b. meyer and h. wittig , pos confinement x , 112 ( 2012 ) . k. tuchin , adv . high energy phys . * 2013 * , 490495 ( 2013 ) d. e.kharzeev , l. d. mclerran , and h. j.warringa , nucl . phys . a * 803 * , 227 ( 2008 ) . d. e. kharzeev , ann . ( ny ) * 325 * , 205 ( 2010 ) . k. fukushima , d. e. kharzeev , and h. j. warringa , phys . d * 78 * , 074033 ( 2008 ) . d. kharzeev and a. zhitnitsky , nucl . a * 797 * , 67 ( 2007 ) . d. e. kharzeev and h. j. warringa , phys . d * 80 * , 034028 ( 2009 ) . v. skokov , a. illarionov , and v. toneev , int . j. mod a * 24 * , 5925 ( 2009 ) . bali , f. bruckmann , g. endrodi , z. fodor , s.d . katz , s. krieg , a. schfer , and k.k . szabo , jhep * 1202 * , 044 ( 2012 ) . g. s. bali , f. bruckmann , m. constantinou , m. costa , g. endrodi , s. d. katz , h. panagopoulos and a. schfer , phys . d * 86 * , 094512 ( 2012 ) . w. cassing and e. l. bratkovskaya , nucl . a * 831 * , 215 ( 2009 ) ; phys . c * 78 * , 034919 ( 2008 ) . l. p. kadanoff and g. baym , _ quantum statistical mechanics _ , ( benjamin , new york , 1962 ) . s. juchem , w. cassing , and c. greiner , phys . d * 69 * , 025006 ( 2004 ) ; nucl . a * 743 * , 92 ( 2004 ) . w. cassing and e. l. bratkovskaya , phys . rept . * 308 * , 65 ( 1999 ) . e. l. bratkovskaya and w. cassing , nucl . a * 619 * , 413 ( 1997 ) . w. cassing , nucl . a * 700 * , 618 ( 2002 ) . w. cassing , v. metag , u. mosel , and k. niita , phys . rep . * 188 * , 363 ( 1990 ) . w. cassing , nucl . phys . a * 795 * , 70 ( 2007 ) . w. cassing , nucl . phys . a * 791 * , 365 ( 2007 ) . a. peshier , phys . d * 70 * , 034016 ( 2004 ) ; j. phys . g * 31 * , s371 ( 2005 ) . e. l. bratkovskaya , w. cassing , v. p. konchakovski , and o. linnyk , nucl . a * 856 * , 162 ( 2011 ) . w. cassing , eur . j. st * 168 * , 3 ( 2009 ) . a. amato , g. aarts , c. allton , p. giudice , s. hands , and j .- skullerud , phys . lett . * 111 * , 172001 ( 2013 ) . o. linnyk , w. cassing , j. manninen , e.l . bratkovskaya , p.b . gossiaux , j. aichelin , t. song , c.m . ko , phys . c * 87 * , 014905 ( 2013 ) .	we study the electric conductivity as well as the magnetic response of hot qcd matter at various temperatures @xmath0 and chemical potentials @xmath1 within the off - shell parton - hadron - string dynamics ( phsd ) transport approach for interacting partonic systems in a finite box with periodic boundary conditions . the response of the strongly - interacting system in equilibrium to an external electric field defines the electric conductivity @xmath2 whereas the response to a moderate external magnetic field defines the induced diamagnetic moment @xmath3 ( @xmath4 ) as well as the spin susceptibility @xmath5 . we find a sizeable temperature dependence of the dimensionless ratio @xmath6 well in line with calculations in a relaxation time approach for @xmath7 as well as an increase of @xmath2 with @xmath8 . furthermore , the frequency dependence of the electric conductivity @xmath9 shows a simple functional form well in line with results from the dynamical quasiparticle model ( dqpm ) . the spin susceptibility @xmath10 is found to increase with temperature @xmath0 and to rise @xmath11 , too . the actual values for the magnetic response of the qgp in the temperature range below 250 mev show that the qgp should respond diamagnetically in actual ultra - relativistic heavy - ion collisions since the maximal magnetic fields created in these collisions are smaller than @xmath12 which defines a boundary between diamagnetism and paramagnetism . + pacs : 12.38.mh , 11.30.rd , 25.75.-q , 13.40.-f
a pair of fermions occupying different internal states can interact in s - wave scattering . fermi atoms can be loaded into different internal spin projection states to produce interacting degenerate fermi gas with @xmath1 . comprehensive reviews of the recent theoretical and experimental advances in the study of the dilute fermi gases can be found in the ref . the fermionic alkali atoms can occupy hyperfine states that result from the coupling of the nuclear angular moment @xmath2 with the electronic spin @xmath3 . in the presence of a strong magnetic field , the electronic degree of freedom is polarized and the relevant internal degrees of freedom are determined by the state of the nuclear angular momentum . in the case of @xmath4li , the nuclear angular momentum @xmath5 allows three different projections . by combining magnetic as well as optical trapping fields , constraints on the trappable hyperfine states are largely removed . for a fermi gas at low density , an expansion of the ground state energy in terms of a small parameter @xmath6 is known @xcite . in general , when we have a gas of fermions occupying @xmath7 different states with the same partial densities , we call it the degeneracy @xmath7 fermi gas . the ground state energy has an expansion in powers of @xmath6 = _ f\ { + ( s-1 ) + o[(ak_f)^3 ] } [ eqn_lenz ] where we identify the zeroth order term as the free fermi gas energy @xmath8 . @xmath9 is the fermi energy , @xmath10 the fermi momentum and @xmath11 the s - wave scattering length . @xmath12 is independent of @xmath7 and it is used as the unit of energy throughout this article . the terms that depend on the interaction potential range @xmath13 are to appear at higher orders @xcite . for @xmath14 fermi gas , the dependence on @xmath13 can be eliminated by taking the limit @xmath15 , where @xmath16 is the average inter - particle distance ( @xmath17 with @xmath18 density ) . the _ intermediate regime _ where @xmath19 is of particular interest for @xmath20 fermi gas . this is also called the _ unitarity regime_. in this regime , the mean free path of the atoms @xmath21 becomes much shorter than @xmath16 as the cross section @xmath22 diverges . examples of the systems in this regime are found in the dilute gases of @xmath23li and @xmath24k atoms at feshbach resonances close to the zero temperature . the magnitude of the s - wave scattering length @xmath25 can be @xmath26 at the feshbach resonance while the interaction range of the van der waals forces @xmath27 . another example is that of the neutron gas . the neutron - neutron interactions by strong force have @xmath28 while @xmath29 . at @xmath30 , zero energy two particle bound state appears . however , it is known that the ground state of the many body system has a positive energy per particle @xmath31 @xcite . ground state properties for @xmath14 fermi gas were studied@xcite using the quantum monte carlo for the unitarity as well as other regimes of interaction . the generalization of the low density expansion ( eq . [ eqn_lenz ] ) to @xmath32 becomes troublesome as the limit of @xmath33 can not be taken . this is expected since for the three particle systems the efimov effect predicts an effective three particle attractive interaction @xcite of the form @xmath34 where @xmath35 is a universal constant . thus , the minimum set of parameters to describe the @xmath36 system consist of @xmath6 and @xmath37 . in general , the mean field treatments of the three component fermi gas @xcite do not account correctly the three particle physics . recently , a stable three component degenerate fermi gas @xcite has been created experimentally . broad and close lying feshbach resonances make the strong and attractive interactions among the atoms in the different internal states possible @xcite . for simplicity , we label the atoms occupying these three different states by a coloring scheme : _ green _ ( g ) , _ red _ ( r ) and _ blue _ ( b ) . the channel dependent s - wave scattering lengths ( @xmath38 , @xmath39 , and @xmath40 ) have been measured experimentally @xcite . it is experimentally difficult to achieve simultaneously strong interactions in all of the channels . however , in the present work we assume a simplified su(3 ) symmetric model where @xmath41 . also , the mass is assumed to be the same for all the components . we leave non - su(3 ) symmetric cases to future study . three component fermi gas problem is also relevant in relation to the _ color superconductivity _ of the quark matter @xcite . in this article , _ ab initio _ monte carlo results of the three component fermi gas are presented . the fermions interact pairwise , by a short but finite range attractive interaction potential . the possibility of the su(2)@xmath0u(1 ) broken symmetry ground state @xcite is considered . here , only two components participate in the superfluid pairing while the third component remains in the normal phase . at weak interaction strengths , the superfluid pairing is exponentially suppressed with the quasiparticle gap @xmath42 . however , we find that there is an interval of the interaction strength where the broken symmetry pairing has noticeable effects in the ground state energy , the quasiparticle gap , the pair distribution functions , etc . when the strength of the interaction is further increased , the su(3 ) symmetry is restored by the predominant three body effects . we present our analysis using the dimensional arguments for the three particle system ( sec . [ sec_bound ] ) and the quantum monte carlo method for the three component fermi gas ( sec . [ sec_many],[sec_results ] ) . thomas @xcite noted that in the nucleus of tritium ( @xmath43h ) which has two neutrons and one proton , the binding energy has no lower bound if we assumed finite negative s - wave scattering length for the proton - neutron interaction and took the interaction range to zero . thus , three particle binding energies depend on the range of the potential and the ground state energy diverges in the limit of zero interaction range . in addition , as the pairwise interaction approaches the resonance ( @xmath44 ) , infinite number of shallow three particle bound states ( called trimers or trions ) appear one after another . they are known as the efimov states @xcite . this property is strikingly different from the two particle case . consequently , qualitatively different behavior of the three component gas is expected to emerge in comparison with the two component fermi gas . for our discussion , we consider a generic three particle hamiltonian where the particles interact pairwise _ 3 = - _ i \{g , r , b } _ i^2 + v_0_i < j \{g , r , b } v_r(r_ij ) . [ eqn_3body_h ] there are two positive parameters in the hamiltonian : strength @xmath45 and range @xmath13 of the interaction . we assume a negative dimensionless function @xmath46 while @xmath47 has dimension @xmath48 . @xmath49 is solely parametrized by @xmath13 . the s - wave scattering length @xmath11 can be used equivalently instead of the strength @xmath45 . in the two particle systems the bound state threshold is at the resonance ( @xmath50 ) independent of the finite @xmath13 . given a value of @xmath13 , it is easy to see that three particle threshold @xmath51 can be such that @xmath52 but not equal to @xmath53 . this can be done using an ansatz as the trial wave function ; for example , a function of the form @xmath54 with the variational @xmath55 . in this case , it suffices to provide a trimer state upper bound in energy to prove the existence of a trimer state with @xmath51 in the interval @xmath56 for a given @xmath13 . the fact that @xmath51 is negative and @xmath57 is the starting point for the analysis on the scaling behavior of the length unit . when we consider rescaling of the length by taking @xmath58 and @xmath59 ( we will always assume @xmath60 from now and on ) , consistent scaling behavior with @xmath61 is expected ( we will justify in the next paragraph ) . here , it can be seen easily from the zero energy scattering solution that the s - wave scattering length @xmath11 scales analogous to another length quantity @xmath13 . by making @xmath62 we expect @xmath63 , @xmath64 and @xmath65 for any @xmath11 that belongs to the interval @xmath66 . this means that , for a zero range interaction , the trimer state is possible for any attractive pairwise interaction of nonzero strength . now , we consider formally the scaling behavior of the energy for the system of three particles in vacuum . we assume a value of @xmath13 such that trimer state is allowed with a strength @xmath67 corresponding to @xmath68 . the three particle schrdinger equation with the usual notation @xmath69 and @xmath70 is _ 3 ( * x * ) = e ( * x * ) . [ eqn_3body_sch ] let @xmath71 , @xmath72 , @xmath73 , @xmath74 and @xmath75 . then @xmath76 and @xmath77 . thus , after scaling all the length quantities by @xmath78 , the eq . [ eqn_3body_sch ] becomes & & _ s(x_s ) + & = & ( * x * ) + & = & e ( * x)= e_s _ s(x*_s ) . as a result , after length scaling @xmath79 is the solution with the eigenvalue @xmath80 . we arrived at this property by using the dimensional arguments alone . as good illustrative examples of this scaling behavior , the following cases are mentioned : when @xmath81 for a given @xmath13 , there is trimer state with energy @xmath82 . then , we can scale length by an overall factor @xmath78 but keep @xmath83 constant ( that is , increase @xmath84 to match @xmath25 ) . it is obvious that taking @xmath62 causes @xmath85 to collapse rapidly to @xmath53 . this means that given a trimer state in vacuum , when @xmath33 at a fixed non - zero value of @xmath11 ( @xmath86 ) , @xmath87 has to go to @xmath53 . this is in agreement with the above mentioned ref . @xcite . as another example , if we had initially @xmath88 , the interaction potential is not strong enough to allow trimer state and the total energy of the particles @xmath89 in the vacuum . in this case , length scaling leaves the particles unbound with @xmath90 . thus , unbound particles remain unbound even after length scaling and the collapse does not occur . for the comparison purpose , let s also consider the scaling behavior of a pair of particles . we can see that the scaling of length does not produce the collapse as in the three particle case . for the s - wave scattering length @xmath11 such that @xmath91 , there is no bound state and the energy of the pair remains zero ( @xmath89 ) in the vacuum at any length scale . then , we consider @xmath92 regime ( usually called bec regime ) . using the same scaling analysis , we also have @xmath93 . in this case , we can solve exactly the contact interaction ( @xmath94 ) problem by replacing the potential by the boundary condition @xmath95 where @xmath96 is the radial wave function of the pair . the solution for the radial wave function is @xmath97 with @xmath98 . this energy is finite unless @xmath99 . according to this solution , the scaling behavior of energy @xmath100 is the same as the result we obtained from the dimensional analysis , @xmath101 . thus , for a pair the energy scaling relation becomes identical to the bound state energy in the strong coupling limit @xmath102 . this is qualitatively different from the three body collapse at the weak coupling limit @xmath103 because of the simultaneous scaling of @xmath13 to zero . for the study of many body systems at finite density , we use _ ab initio _ stochastic method known as fixed node green s function monte carlo ( fn gfmc ) . in general , we take a trial wave function @xmath104 that obeys antisymmetry upon the exchange of identical fermions and let it evolve in the imaginary time restricted to a definite sign domain given by the nodal surface of the trial wave function itself . if the nodal structure is correct , we get the exact ground state . otherwise , we get an approximate ground state and an energy upper - bound . the implementation of this method is explained in detail elsewhere @xcite . we consider a system with @xmath105 particles of each color ( green , red , blue ) , so that @xmath106 . the particles are contained in a finite box with the periodic boundary conditions at the walls to simulate the uniform matter . for many particle systems , we can not rescale the length without changing the density . the scaling behavior analyzed in the previous section is only applicable to the few particle systems in vacuum . in the degeneracy two case , one dimensionless product @xmath107 uniquely determines the system . the parameter @xmath13 can be pushed in principle to the zero limit and eliminated from the description of the system . in practice , small but finite values of @xmath108 were assumed @xcite as long as the results were converged within the statistical errors to the @xmath109 limit . however , from the scaling behavior analysis of the previous section , it becomes clear that for the degeneracy three fermi gas we need to keep @xmath37 finite in order to avoid local trimer instability . thus , we need both @xmath6 and @xmath37 to fully describe the three component fermi gas . we keep @xmath110 in the @xmath111 case which is the same value as in the @xmath112 case ( @xmath113 , thus converged to the @xmath15 limit ) . however , we should keep in mind that for @xmath36 , this particular value of @xmath13 is not the limit of @xmath33 . we consider the cases with small deviations from the balanced partial densities . we explore the possibility of the superfluid ground state within a certain interval of the interaction strength . we also analyze the qualitative behavior of the pair distribution functions ( see sec.[sec_results ] ) . the many body su(3 ) symmetric hamiltonian is = - _ i _ i^2 + v_0 _ i < j v_r(r_ij)(1-_c_i , c_j ) where @xmath114 is the color of i - th particle . only pairs of different color particles interact . @xmath49 is the dimensionless core of the pschl - teller potential : @xmath115 and @xmath116 . @xmath45 can be adjusted to get the desired scattering lengths . we can impose different nodal restrictions to the solution by using different trial wave functions . we can estimate the energy by using su(3 ) symmetric slater trial wave function _ fg= _ fg , g_fg , r_fg , b [ eqn_fg ] where the factors represent the normal states ( given by slater determinants of the plane wave orbitals ) of different color species . it was also shown@xcite that the pairing fields @xmath117 with @xmath118 can be mapped into @xmath119 with the constraint @xmath120 . this is analogous to the analysis of ref . @xcite by modawi and leggett where the ground state should allow one normal phase component . thus , we consider the su(2)@xmath0 u(1 ) broken symmetry pairing ground state with bcs pairing for two of the fermi components while the third component remains in the normal phase . the corresponding nodal structure is given by the trial wave function _ bs - bcs & = & _ fg , b _ bcs , gr + & = & \|0 + & & _ fg , b a[(r_11 ) ( r_22 ) ... ( r_mm)]_gr . [ eqn_bs_bcs ] here , we assume that blue species remains in the normal phase ( represented by @xmath121 ) , while between the green and red species there is pairing correlation ( represented by @xmath122 ) . in the last line of eq . [ eqn_bs_bcs ] , we assumed fixed number projection of the green and red particles . the complete trial wave function with jastrow - like factor can be written as _ v , fg = _ i < j_fg [ eqn_res3_slater ] for the su(3 ) symmetric slater wave function ( eq . [ eqn_fg ] ) . here , we considered separately the correlation that exists between the same color particles @xmath123 ( @xmath124 ) from the correlation that exists between different color particles @xmath125 ( @xmath126 ) . usually @xmath123 is qualitatively different from @xmath125 . @xmath123 is analogous to @xmath127 of the degeneracy two case and includes pauli exclusion principle ( @xmath128 ) . the particular shapes of the correlation functions do not change the gfmc energies . however , they are optimized in order to minimize statistical errors and to have the converged pair distribution functions @xmath129 ( see ref . we can see that the optimized @xmath130 deviates largely ( more peaked at @xmath131 ) from the one obtained by using the locv equations @xcite . this is due to the strong three body effects even when the pairwise interactions are relatively weak . analogously , the complete trial wave function with the su(2)@xmath0u(1 ) broken symmetry pairing correlation ( eq . [ eqn_bs_bcs ] ) is _ v , bs - bcs & = & _ i < j [ f_s(r_ij)_c_i , c_j + f_gr(r_ij)_c_i , g_c_j , r + + & & f_gb(r_ij)(_c_i , g + _ c_i , r)_c_j , b ] _ bs - bcs . [ eqn_res3_bcs ] this wave function can give better pair distribution functions as the optimization of the correlation functions can be carried out separately for @xmath132 and @xmath133 . then , we can demonstrate that @xmath134 as well as @xmath135 ( see the discussion of the fig . [ fig_two ] in the sec . [ sec_results ] ) . the optimum nodal structure is tried as @xmath136 with non zero short range function @xmath137 parametrized by @xmath138 ( see ref . @xcite for the definitions of these parameters ) . these parameters are identical to those of the @xmath20 case when @xmath139 ( called molecular or bec regime ) . this trial wave function assumes arbitrarily one of the three possible broken symmetry pairing states \|1 & & _ fg , b _ bcs , gr + and the broken symmetry can be assumed by choosing one of these states without loss of generality . the gfmc energies using the trial nodes given by the wave functions of eq . [ eqn_res3_slater ] and eq . [ eqn_res3_bcs ] are summarized in the fig . [ fig_one ] , while the pair correlation functions are presented in the fig . [ fig_two ] . ) and three ( @xmath111 ) results for @xmath140 in both cases . @xmath141 ( eq . [ eqn_lenz ] ) gives a good estimate of the normal state energy for @xmath14 and @xmath142 . however for @xmath111 , the match is poor with the gfmc results as the @xmath13 dependent terms and three particle effects predominate . in the region @xmath143 , the su(2)@xmath0u(1 ) broken symmetry pairing state is shown to be the ground state for @xmath111 fermi gas . however , for @xmath144 the su(3 ) symmetry is restored(see also fig . [ fig_four ] ) and the superfluid pairing suppressed . this behavior for @xmath36 is quantitatively different from the @xmath20 case , where with the increasing interaction strength ( increasing @xmath145 ) , the superfluid pairing state ( crosses ) is always the favored ground state compared to the state without superfluid correlation ( circles ) . the size of the symbols approximately correspond to the error bars . , width=302 ] ( a ) , @xmath146(b ) , @xmath147(c ) and @xmath148(d ) . we define pair distributions @xmath149 . the angled brackets imply taking thermal average of the possible configurations . @xmath150 has boundary condition @xmath151 . the line corresponds to @xmath152 , that is , the pair distribution between the green and red species while the dashed line corresponds to the distribution between the green and blue species @xmath153 . both @xmath152 and @xmath153 assume su(2)@xmath0u(1 ) symmetric state ( eq . [ eqn_res3_bcs ] ) . the dotted line represents the pair distribution without superfluidity @xmath154 calculated with the su(3 ) symmetric slater wave function ( eq . [ eqn_res3_slater ] ) . the @xmath154 is independent of the colors as long as they are different . all @xmath155 s are calculated with the optimized @xmath55 functions . in the plots c ) and d ) where the superfluid pairing is strong , @xmath156 . thus , on the average the particles that have superfluid pairing correlation remain further apart than the particles that have no superfluid pairing correlations . , width=302 ] . the circles correspond to the set of particle numbers @xmath157 @xmath158,@xmath159,@xmath160 , @xmath161,@xmath162,@xmath163 , and @xmath164 . the squares are @xmath165 , @xmath166 , @xmath167 , and @xmath168 cases . the triangles represent the configurations @xmath169 and @xmath170 . the size of the symbols approximately correspond to the error bars.,width=226 ] are shown . each time step is @xmath171 . the states @xmath172 , @xmath173 and @xmath174 are as defined in eq . [ eqn_degens ] while @xmath175 is as defined in eq . [ eqn_res3_slater ] . we can see that as the interaction strength is increased , the symmetry is restored with the overlaps approaching 1.,width=302 ] the results of three component fermi gas energies are summarized in the fig . [ fig_one ] . the pairing correlations produce noticeable effects at @xmath176 ( compare the triangles with the circles of the same figure ) . close to the @xmath177 , the pairing effects are small . here , the energy of the su(2)@xmath0u(1 ) broken symmetry state is not distinguishable within the error bars from that of the su(3 ) symmetric state ( slater ) . the system is found stable in the regime of interaction considered in this work ( @xmath178 and @xmath140 ) . we also found that it is possible to see the effects of pairing in the @xmath155 s . in the plot a ) of fig . [ fig_two ] ( @xmath179 ) , no difference can be seen in the @xmath155 s because the pairing is weak . however , where the pairing is relevant ( in terms of the energy ) we notice that @xmath180 ( where @xmath154 is the pair distribution obtained with the non - pairing slater wave function ) and the symmetry is broken ( see the plots c ) and d ) of the fig [ fig_two ] ) . green and red particles have superfluid pairing correlations , so one may have naively expected @xmath181 , but the opposite is found to be true . according to this , the relative distances satisfy on average : @xmath182 . the interpretation we can give is that once a cooper pair is formed , the third particle feels much stronger attraction toward the center of the mass of the cooper pair . in fact , we can approximately estimate that the potential strength between the center of mass ( cm ) of green - red pair and the blue particle is enhanced from @xmath45 to @xmath183 , while green and red particles interact with the strength @xmath45 . this comes from the observation that the zero energy scattering equation between the cm of green - red and blue is @xmath184 with @xmath185 . from the plot d ) of the fig . [ fig_two ] , it is still not conclusive whether the symmetry could be restored ( @xmath186 ) in the regime of strong three particle correlation @xmath187 . in that regime , the analysis is hindered on the practical ground : it becomes increasingly harder to obtain reliable estimates of @xmath155 keeping the statistical errors small . however , the fig . [ fig_one ] and fig . [ fig_four ] show clearly the tendency toward the restoration of the su(3 ) symmetry entering into a regime completely dominated by three particle physics with negligible paring correlations . using the gfmc technique , energy gaps can be calculated by allowing variations in the numbers of the green and red particles while keeping the number of blue particles constant . in this way , we allow breaking of the superfluid pairs . we can consider sets of @xmath157 @xmath158,@xmath159,@xmath160 , and @xmath161,@xmath162,@xmath163 . we observe the usual odd - even staggering of the ground state energy ( fig . [ fig_three ] ) . furthermore , we can allow the variations @xmath188 , @xmath189 , and @xmath190 and calculate the excitation energies . the lowest energy excitations at @xmath191 and @xmath147 are found with the momentum @xmath192 quasiparticle . in the case of @xmath14 , quasiparticles with zero momentum produce the minimum excitation in the @xmath193 ( bec ) regime . thus , this is consistent with the interpretation that the studied regime ( @xmath194 ) is that of strong three particle correlation . the energies for the broken pair states are shown in the fig . [ fig_three ] for the @xmath195 case . the gap @xmath196 is estimated from the data sets represented by circles with total @xmath197 and @xmath198 respectively . the calculated @xmath199 . the error bars are large for the pairing gap since the three particle effects predominates rather than two particle pairing . in the fig . [ fig_three ] , we notice that the energy dips when total @xmath200 is a multiple of @xmath201 . we interpret this as an effect analogous to that observed in the @xmath155 s . this is the evidence that the trimer interaction that brings the green - red pair and the blue particle together is much stronger than simple pairwise interaction . thus , completing green - red - blue trimer is energetically more favorable than unbalanced excess of one or two species . in fact , the trimer binding energy is so strong that @xmath202 in contrast to @xmath203 of @xmath20 fermi gas in the bec regime . consequently , in the quasiparticle spectrum , it is expected that we can observe two distinctive _ gaps _ ; one due to the superfluid pairing and another due to the trimer binding . we also notice that at @xmath204 the dip is as deep as at @xmath205 which indicates possible shell closure effect . as seen in the ground state energy ( fig . [ fig_one ] ) and the overlap of the wave functions ( fig . [ fig_four ] ) , the broken symmetry superfluidity is suppressed in the @xmath206 regime where the su(3 ) symmetry is restored . for the comparison purpose only , we _ naively _ consider extension of the mean field ( bcs - leggett ) method to the three component fermi gas . for this mean field model , there is no @xmath13 dependence and the variational ground state is always stable . the two color pairing gap of the degeneracy three superfluid is given by the same relation as that of the degeneracy two fermi gas . while in the original bcs formalism , the chemical potential @xmath207 is kept constant and @xmath208 we consider the bcs - leggett @xcite variational formalism ( @xmath209 ) . here , the chemical potential is changed in order to keep the density constant . the condensation energy of the system is @xmath210 where @xmath211 is the state density ( @xmath212 ) . dependence on the degeneracy @xmath7 is included in the condensation energy per particle @xmath213 because of the relation @xmath214 . thus , at @xmath215 , we estimate that @xmath216 and @xmath217 . at @xmath218 ( or @xmath219 ) , we have @xmath220 and @xmath221 . this estimate is close to the one calculated by gfmc ( fig . [ fig_three ] , @xmath222 ) at the same interaction strength . although for @xmath36 fermi gas , @xmath13 dependence can not be removed , clear qualitative differences between the @xmath14 and @xmath36 fermi gases emerge . unlike in the @xmath20 fermi gas where both the superfluid pair and the bound state are qualitatively similar , in the @xmath36 fermi gas the paired states decouple from the bound state(trimer ) in energy . this can be clearly observed in the quasiparticle excitation spectrum . realistic interaction potential and channel dependence of the interactions are necessary in order to produce not only qualitative but also quantitatively correct results for a given three component fermi gas . we found a regime of interaction strength where the broken symmetry pairing is clearly detectable , beyond which three particle effects dominate and the symmetry is restored . also , non trivial dependence of the @xmath155 s on the pairing correlations was discussed . this work has been supported in part by the us national science foundation via grant phy 00 - 98353 and phy 03 - 55014 . syc also acknowledges support by the darpa grant baa 06 - 19 . the authors acknowledge useful comments from j. carlson and a. bulgac . one of the authors(vrp ) passed away during the preparation of the present manuscript and the work was posthumously completed .	we present an analysis of the su(3 ) symmetric model of the strongly interacting three component fermi gas in the continuum space using quantum monte carlo techniques . three body effects predominate in the regime of interaction strength beyond that of threshold of the three particle bound state . however , we find that there is an interval of the interaction strength where the su(2)@xmath0u(1 ) broken symmetry superfluidity is possible . for a strong enough interaction , the su(3 ) symmetry is restored and the superfluidity is suppressed . within the interval of the broken symmetry , we also find that on average the particle pairs belonging to the species with superfluid pairing remain further separated than those without the superfluid pairing correlation .
cold atomic gases give a possibility to address collective atomic quantum states , where an elementary atomic excitation is coherently distributed among a large number of individual atoms . such systems have been considered in the context of quantum information @xcite , coherence protection @xcite , individual photon manipulations @xcite etc . for cold rydberg gases that have large dipole moments , one has to allow for the effects associated with strong dipole - dipole interaction among atoms @xcite-@xcite . considering the regime typical of current experimental setting , one usually takes into account the regular static dipole - dipole interaction @xmath0 among pairs of the atoms , which from the view point of quantum electrodynamics is a result of exchange by a virtual strongly off - resonant vacuum photon with a typical wavelength of the order of interatomic distance @xmath1 @xcite . this sort of interatomic coupling can be attributed neither to the short range interaction , nor to the long range ones , since the integral @xmath2 corresponding to the average binary interaction logarithmically diverges at both upper and lower limits . this circumstance results in a number of interesting dynamic phenomena @xcite , such as incomplete decay of a single atom population at the limit of long times , which resembles the anderson localization effect . in the same time , for the description one can not make use of the continuous media model and the mean field approximation , which are applicable only for the long range interactions . among experiments with rydberg atoms , there are some @xcite that have been done in the regime where the size of a cloud of cold rydberg atoms is of the order or larger than the wavelength of the resonant dipole active transition . in such conditions , the role of the radiation trapping , super - radiance @xcite , and the dynamic long - range dipole - dipole interactions ( @xmath3 ) is dominant , and the continuous media model becomes applicable . in other words quantum states of the atomic ensemble and that of the resonant radiation field get strongly entangled , whereas the eigenstates of the compound atoms+field system correspond to atomic states dressed in the resonant radiation field . driven by the curiosity we will now consider the atomic population dynamics in the simplest version of such a process . we make use of the mean - field approximation , and concentrate on an exactly soluble case of an infinite uniform continuous two - level media , ignoring the contribution of the static dipole - dipole interaction not conforming the requirements of the mean field model . we consider a static continuous media corresponding to the gas of @xmath4 two - level particles with the transition frequency @xmath5 and polarization @xmath6 along @xmath7-direction that are uniformly scattered in space at fixed positions . since the static dipole - dipole interaction between the atoms can be viewed as a process intermediated by virtual photon exchange , the effective hamiltonian that governs the evolution of the combined system in the rotating wave approximation , is @xmath8 where @xmath9 are the collective operators of the exciton @xmath10 , @xmath11 is the coupling and @xmath12 is the angle between the direction of the polarization @xmath6 and the normal to the photon polarization plane . the coupling strength @xmath13 is a product of the photon vacuum field strength @xmath14 in a volume @xmath15 and the collective atomic dipole moment @xmath16 , where @xmath17 is the atomic transition dipole moment , @xmath18 is the speed of light , and @xmath19 . we assume that at time @xmath20 , all but one particles are in the ground state . the consideration is equally applicable to the case where a group of particles locating in a volume with a typical size @xmath21 , small as compared to the resonant transition wavelength @xmath22 , is in a coherent superposition of the individual excited states , such that the total number of excitations is one . in order to take advantage of the uniform distribution , we take states@xmath23 corresponding to a given wavevector @xmath10 as the basis set in the collective hilbert space . here @xmath24 denotes the vacuum where all particles are in the ground state , and the condition that the average distance @xmath25 among the neighboring particles corresponding to the density @xmath26 is much shorter than the transition wavelength @xmath27 is implicit . each state , ( [ a ] ) , interacts only with the photon of the same wavevector , and the amplitudes @xmath28 and @xmath29 of the particle+field compound state @xmath30 ( where the second quantum number corresponds to photons ) satisfy the schrdinger equation@xmath31 where dirac delta function @xmath32 stands for the initial condition corresponding to the excitation location in the origin , that is at the point @xmath33 . the exact solution of ( [ ba ] ) and ( [ bb ] ) @xmath34 + \frac{(\omega - kc)\sin\left [ \omega t\right ] } { 2\omega}}{\left ( 2\pi\right ) ^{3}}\\ \varphi_{\mathbf{k } } = & -\frac{u\sqrt{k}\sin\left ( \alpha\right ) } { \left ( 2\pi\right ) ^{3}}\mathrm{e}^{\mathrm{i}t\frac{\omega - kc}{2}}\frac{\sin\left [ \omega t\right ] } { \omega } \label{c}\ ] ] with the rabi frequency @xmath35 suggests @xmath36 + \frac{(\omega - kc)\sin\left [ \omega t\right ] } { 2\omega}}{-\left ( 2\pi\right ) ^{3}}\mathrm{d}\gamma\label{da}\\ \varphi(\mathbf{r } ) & = \int\frac{k^{2}\mathrm{e}^{-\mathrm{i}kr\cos \theta+\mathrm{i}t\frac{\omega - kc}{2}}}{-8\pi^{3}}u\sqrt{k}\sin\left ( \alpha\right ) \frac{\sin\left [ \omega t\right ] } { \omega}\mathrm{d}\gamma\label{db}\ ] ] for the coordinate dependent amplitudes , with @xmath37 , where @xmath38 denotes the integration over solid angle and the integration over @xmath39 starts at the point @xmath40 and goes to @xmath41 . long range dipole - dipole interaction corresponds to the exchange by photons close to the resonance domain @xmath42 . in this domain , the integrands ( [ da])-([db ] ) have two branching points @xmath43 in the complex plane of @xmath39 where @xmath44 assumes zero value . therefore the contribution of the long range interaction is given by an integral along a contour presented in figure [ contour ] circumventing these points , which can be expressed in terms of the bessel functions @xmath45 . in order to shorten the notations we employ the units where @xmath46 henceforth . in the limit @xmath47 of a coupling , which is weak as compared to the energy of atomic transition quantum , the integration over @xmath48 yields@xmath49 up to terms of higher orders in @xmath50 . here @xmath51 is the heaviside-@xmath52 functions , while @xmath53 and @xmath54 stand for @xmath55 and @xmath56 , respectively . integration over the solid angle in ( [ ea])-([eb ] ) has to be done numerically in the polar coordinates associated either with the direction of the polarization @xmath6 ( angles @xmath12 and @xmath57 ) or with the direction of the radius vector @xmath58 ( angles @xmath52 and @xmath59 ) as shown in figure [ fig.1 ] . it has to be performed over the intervals @xmath60 and @xmath61 , respectively . angles @xmath12 and @xmath52 do not belong to a same reference frame and therefore one of these angles has to be expressed in the frame , to which belongs the other one . for the purpose , one can make use of the relations@xmath62 that follow from the invariance of the scalar product @xmath63 . alternatively one may employ the cylindrical set of coordinates @xmath64 , @xmath65 , and @xmath57 , associated with the direction of the polarization @xmath6 . we note that there are two size scales in the problem : the wavelength @xmath66 ( that is @xmath67 in the dimensional units ) and a distance @xmath68 ( that is @xmath69 in the dimensional units ) at which the light propagates during the rabi period at resonance . population distribution at these scales have to be analyzed separately . let us first consider the wavelength scale . for short times @xmath70 , the bessel function can be cast in taylor series and the first order yields @xmath71 for the amplitude in ( [ eb ] ) . with the help of equation ( [ fa ] ) one can calculate the integral exactly and find@xmath72 @xmath73which results in the population @xmath74 quadratically increasing in time , shown in figure [ fig2 ] . note , that at long distances the probability corresponding to ( [ h ] ) decreases as @xmath75 , which means that the integral of the population over the volume logarithmically diverges at the upper limit being truncated just at the radius @xmath76 . in other words , some part of the excitation is getting transferred by the resonant photons at a large distance @xmath77 . the distribution of figure [ fig2 ] persists as long as @xmath78 remains a small number . its structure can be easily understood when one notice that according to the second equation ( [ c ] ) , the amplitudes of emitted photons are located in the domain of resonance where the the detunings @xmath79 are small . these photons , after being emitted , result in discarding of the corresponding spacial harmonics from the initial @xmath80-like excitation profile @xmath81 containing all harmonics with equal amplitudes . but such a discarding implies establishing of the population distribution given by the discarded harmonics taken with the opposite signs . as the result , the sum of the @xmath80-like distribution and the latter distributions of @xmath81 no longer contains harmonics strongly coupled to the cooperative resonant radiation . in course the of time the harmonics with higher detuning are getting involved in the process , and the net population of figure [ fig2 ] increases . for an initial distribution of a small but finite size @xmath21 , the relative amount of this population is given by the ratio of the length @xmath82 of the @xmath39-interval of emitted photons and the length @xmath83 of the @xmath39-interval occupied by the initial distribution . in course of time @xmath84 , the distribution @xmath74 in the domain @xmath85 gets modified , as shown in figure [ fig3a ] . one can find an analytic expression for the asymptotic form of @xmath86 at @xmath87 . the main contribution to the integral ( [ eb ] ) comes from the domain of small @xmath88 , that results from the exchange by the resonant photons with @xmath10 almost orthogonal to @xmath58 . integration over @xmath38 can be carried out in the reference system associated with @xmath58 with the allowance for the approximate relation @xmath89 and yields@xmath90 one sees that the population distribution decreases with the radius as @xmath91 and has an angular dependence @xmath92 @xmath93 shown in figure [ fig5 ] . ) of the angular part of the population distribution on time @xmath94 . in the inlet we show the contour plot of the population distribution calculated with the assymptotic formula for @xmath95.,width=269,height=165 ] angular shrinking of the distribution in the course of time can be understood if we note that the photons emitted in the direction close to that of the particles polarization are weekly coupled due to the smallness of the projection of the electric strength vector to the particle dipole moment . therefore the typical evolution time for their interaction with the atomic cloud is longer . for two values of the parameter @xmath96 ( a ) and @xmath97 ( b ) resulting from the numerical integration of equation ( [ eb]).,width=244,height=257 ] what happens at the large scale with typical distances @xmath98 ? the results of numerical integration of ( [ eb ] ) are shown in figure [ fig3 ] for two different times . one sees rather irregular structures to some extend resembling the time dependence of fractional revivals @xcite , which results from the interference of different spacial harmonics with the wavevectors close to unity . one should note that this regime implies that the size of the system is larger or comparable with the distance @xmath98 . however , this distance can be relatively large , exceeding the typical size of the atomic cloud . ) for an atomic cloud of a finite size @xmath99 and for the atomic density profile @xmath100.,width=212,height=164 ] in such a situation , instead of considering the population distribution at distances @xmath101 , one may calculate the frequency spectrum of photons detected at a distance much larger as compared to the typical size of the atomic cloud . the calculations relevant to the case of large distances rely on the fourier transformed ( [ ba ] ) and ( [ bb ] ) @xmath102 while the resulting amplitude@xmath103 to have a photon with the wavevector @xmath10 at an angle @xmath12 to the direction of the polarization has to be averaged over an interval around @xmath104 of a width @xmath105 corresponding to the uncertainty resulting from the finite size @xmath106 of the system . the averaging has to be performed with a weight function @xmath107 given by the fourier transformed particle density profile@xmath108 for an analytic profile @xmath107 this yields the spectral intensity@xmath109 where the frequency deviation from the resonance @xmath110 is scaled by the interaction . in figure [ rad ] we present the probability of the photon detection as a function of frequency calculated for the particle density distribution @xmath111 with @xmath112 . this distribution stands as a physical approximation to the delta function spatial distribution of excited atoms we have assumed throughout the derivation . one sees , that though the cooperative coupling between the particles and the electromagnetic field does not govern the population distribution @xmath74 over the ensemble of two - level particles of a finite size @xmath113 , it manifests itself in the spectrum of emitted photons that can be registered at distances large as compared to @xmath106 . . the population distribution for small times and for long times are shown by the contour plots in the upper and the lower parts inside the circle representing the atomic cloud . in 3d the distribution has cylindrical symmetry around the polarization axis @xmath6 . the process is associated with the cooperative emission of photons , and the typical scale of the spectrum of these photons is of the order of @xmath50.,width=255,height=184 ] we conclude by presenting the overall picture of the process shown in figure [ overall ] . an atomic cloud , initially in the ground state , has been excited in a small domain close to the center . in course of time the excitations gets redistributed over all the volume of the cloud due to the long range dipole - dipole interaction . the typical rate of the process corresponds to the @xmath114 . for short times , the population distribution is given by ( [ h ] ) and for @xmath115 it takes the asymptotic form ( [ i ] ) . the process is associated with the cooperative spontaneous emission of photons . the typical scale corresponding to the spectrum of emitted photons is given by the interaction strength @xmath116 , and at a large distance from the atomic cloud it has a two - peak structure of equation ( [ l ] ) . as a general conclusion , one may state that the presence of two essentially different energy scales ( @xmath50 and @xmath117 ) can be considered as typical manifestation of collective atomic phenomena . note that fractional revivals discussed by i. sh . averbukh and n. f. perelman in [ phys . a * 139 * 449 ] are typical of time evolution of quantum systems with discrete spectra that have a small quadratic inequidistance . here a similar phenomenon occurs in the continuous spectrum , which requires a special and more detailed consideration .	we consider the effect of the long range dipole - dipole interaction on the excitation exchange dynamics of cold two - level atomic gase in the conditions where the size of the atomic cloud is large as compared to the wavelength of the dipole transition . we show that this interaction results in population redistribution across the atomic cloud and in specific spectra of the spontaneous photons emitted at different angles with respect to the direction of atomic polarization .
peter offers to let paul toss a fair coin repeatedly until it lands heads and pays him @xmath5 ducats if this happens on the @xmath6 toss , where @xmath7 . this is the so - called classical st . petersburg game . if @xmath8 denotes paul s winning , then @xmath9 , @xmath10 . put @xmath11 for the lower integer part , @xmath12 for the upper integer part , and @xmath13 for the fractional part of @xmath14 . then the distribution function of the gain is @xmath15 and its quantile function @xmath16 is @xmath17 let @xmath18 be i.i.d . st . petersburg random variables , and let @xmath19 denote their partial sum and their maximum , respectively . to define the @xmath4-trimmed sum , let @xmath20 be the ordered sample of the variables @xmath21 . for @xmath22 put @xmath23 that is , from the partial sum we subtract the @xmath4 largest observations . note that @xmath24 is the st . petersburg sum , while @xmath25 is the 1-trimmed sum . in order to state the necessary and sufficient condition for the existence of the limit , we introduce the positional parameter @xmath26,\ ] ] which shows the position of @xmath2 between two consecutive powers of 2 . since the function @xmath27 in the numerator in ( [ eq : dist - stp ] ) is not slowly varying at infinity , the st . petersburg distribution is not in the domain of attraction of any stable law or max - stable law , so limit distribution neither for the centered and normed sum , nor for the centered and normed maximum , holds true . what holds instead for the sum is the merging theorem @xmath28 shown by csrg @xcite , where @xmath29 is the distribution function of the infinitely divisible random variable @xmath30 , @xmath31 $ ] , with characteristic function @xmath32 + \int_0^{\infty } \left ( e^{{\mathfrak{i}}t x } - 1 - \frac{{\mathfrak{i}}t x } { 1+x^2 } \right ) { \mathrm{d}}r_{\gamma}(x ) \right)\ ] ] with @xmath33 , @xmath34 , and right - hand - side lvy function @xmath35 convergence of @xmath36 along subsequences @xmath37 , @xmath38 $ ] , was first shown by martin - lf in 1985 @xcite . for the maximum we have @xmath39 in particular @xmath40 , for any @xmath41 , as @xmath42 , where @xmath43.\ ] ] see formula ( 4 ) by berkes et al . in @xcite , or lemma 1 by fukker et al . in @xcite in the general case . the limit theorems ( [ eq : sum - merge ] ) and ( [ eq : max - pr - merge ] ) suggest that the irregular oscillating behavior is due to the maximum , which is also indicated by the following fact . it is well - known ( see chow and robbins @xcite and adler @xcite ) that @xmath44 while the trimmed sum has nicer behavior , concerning at least the almost sure limits , since @xmath45 ( cf . csrg and simons @xcite ) . for further results and history of st . petersburg games see csrg @xcite and the references therein . as a continuation of our studies of the joint behavior of @xmath46 and @xmath47 in @xcite , we investigate the properties of the trimmed sum @xmath3 both for fix @xmath2 and for @xmath42 . figure [ fig1 ] shows the histograms of the st . petersburg sum and of the 1-trimmed st . petersburg sum . one can see that the histogram of @xmath48 is mixtures of unimodal densities such that the first lobe is a mixture of overlapping densities , while the side - lobes have disjoint support . for the histogram of @xmath49 the side - lobes almost disappear , so the trimmed version has smaller tail . according to proposition 7 in [ 9 ] , for large @xmath47 one gets @xmath50 , or equivalently @xmath51 , which explains the disappearance of side - lobes . and of @xmath49 for @xmath52.,width=377 ] in section [ sect : fixn ] we investigate asymptotic behavior of the tail of the distribution function of the sum @xmath3 for fix @xmath2 . in theorem [ th : snr - tail ] we determine the exact tail behavior of @xmath0 . in particular , we show that the st . petersburg distribution is almost subexponential in a well - defined sense . in section [ sect : ntoinf ] we let @xmath42 . in theorem [ th : trimmed ] we determine @xmath53 \}$ ] , the set of the possible subsequential limit distributions of @xmath54 , and we obtain an infinite series representation for the distribution function . this result was first obtained by gut and martin - lf in their theorem 6.1 in @xcite . they investigate the so - called max - trimmed st . petersburg game , where from the sum @xmath46 all the maximum values are subtracted . theorem [ th : trimlim ] states the limit theorem for the trimmed sums under arbitrary trimming , while in theorem [ th : trimtail ] we determine the tail behavior of the limit . in particular , we obtain exact tail asymptotics for a semistable law . finally , in section [ sect : general ] we mention some of these results without proof in case of generalized st . petersburg games . in this section the number of summands @xmath2 is fix , and we are interested in the tail behavior of @xmath3 . first we summarize some basic facts on subexponential distributions . let @xmath55 be a distribution function of a non - negative random variable @xmath56 . put @xmath57 . the distribution @xmath55 is _ subexponential _ , @xmath58 , if @xmath59 where @xmath60 stands for the usual convolution , and @xmath61 is the @xmath62 convolution power , for @xmath63 . the characterizing property of the subexponential distributions is that the sum of i.i.d . random variables behaves like the maximum of these variables , that is for any @xmath64 @xmath65 or equivalently @xmath66 for properties of subexponential distributions and their use in practice we refer to the survey paper by goldie and klppelberg @xcite . it is well - known that distributions with regularly varying tails are subexponential . what makes the st . petersburg game so interesting is that its tail is not regularly varying . in fact it was already noted by goldie @xcite that the st . petersburg distribution @xmath67 is not subexponential . what we have instead is that @xmath68 this can be proved by showing that for @xmath69 @xmath70 from which @xmath71 moreover , it is simple to see that 4 is in fact the limsup . this naturally leads to the extension of subexponentiality . a distribution @xmath55 is _ o - subexponential _ , @xmath72 , if @xmath73 it is known that the corresponding @xmath74 is always greater than , or equal to @xmath75 , and it was shown recently by foss and korshunov @xcite that it is exactly 2 for any heavy - tailed distribution . the notion of o - subexponentiality was introduced by klppelberg @xcite . the properties of the @xmath76 class , in particular when the distribution is also infinitely divisible , were investigated by shimura and watanabe @xcite . in their proposition 2.4 they prove that if @xmath72 then for every @xmath77 there is a @xmath78 such that for all @xmath2 and @xmath79 @xmath80 in the st . petersburg case @xmath81 . in theorem [ th : snr - tail ] we determine the exact asymptotic behavior of @xmath82 , which , in particular , implies a linear bound in @xmath2 instead of the exponential . let us examine the case @xmath83 in detail . note that @xmath84 for @xmath85 , therefore from ( [ eq : sumdf2 ] ) @xmath86 from this it is clear that when both @xmath87 and @xmath88 tends to infinity , then the limit exists and equal to 2 ; in particular for any @xmath89 @xmath90 where @xmath13 stands for the fractional part of @xmath14 . that is , the st . petersburg distribution is ` almost subexponential ' . we prove the corresponding result for general @xmath2 , i.e. for any @xmath89 @xmath91 [ th : snr - tail ] for any @xmath92 we have as @xmath1 @xmath93 in particular , for any @xmath94 , @xmath95 let @xmath20 be the ordered sample of the variables @xmath21 . using the well - known quantile representation , the form of the quantile function ( [ eq : quant ] ) , and that @xmath96 , for @xmath97 , we obtain @xmath98 where @xmath99 is the ordered sample of @xmath2 independent uniform@xmath100 random variables . introducing the function @xmath101 , i.e. it grows linearly from 1 to 2 on each interval @xmath102 , @xmath103 , we have @xmath104 in the following we frequently use the simple facts @xmath105 and @xmath106 if and only if @xmath107 . the density function of @xmath108 is @xmath109 , therefore @xmath110 considering the asymptotics , write @xmath111 for @xmath112 we have @xmath113 , thus , by ( [ eq : i3 ] ) the first sum @xmath114 for @xmath115 we have @xmath116 here we used the simple fact that conditioning on @xmath117 @xmath118 combining ( [ eq : i2 ] ) and ( [ eq : i3 ] ) formula ( [ eq : exacttail ] ) follows . to show ( [ eq : tail - as ] ) notice that if @xmath119 and @xmath1 , then @xmath120 , which implies the convergence @xmath121 . when @xmath122 the result describes the tail behavior of the untrimmed sum @xmath46 . in figure [ fig : df ] the oscillatory behavior of @xmath123 is clearly visible . we also see that at each power of 2 there is a large jump , that is where the asymptotic ( [ eq : tail - as ] ) fails . in a logarithmic scale.,width=377 ] we mention some important consequences . in the untrimmed case theorem [ th : snr - tail ] gives @xmath124 which readily implies that for any @xmath64 we have @xmath125 since @xmath126 , @xmath127 , we have @xmath128 this convergence also shows that ( [ eq : tail - as ] ) does not hold without the restriction , since by ( [ eq : subexp - prop2 ] ) that would imply the subexponentiality of @xmath129 . for @xmath130 fixed as @xmath131 @xmath132 therefore from ( [ eq : exacttail ] ) we obtain that for any @xmath133 @xmath134 as @xmath131 . for the maximum term we have @xmath135 , and so ( [ eq : finer - as ] ) with @xmath136 gives @xmath137 if @xmath138 tends to infinity arbitrarily slowly , then the limit above is 1 , that is the st . petersburg distribution is very close to having the subexponential property . in the following we determine the possible limit distributions of the 1-trimmed sum , and we investigate the limit . first we introduce some notation . given that @xmath139 for @xmath140 we have @xmath141 . introduce the corresponding distribution function @xmath142 , & \text{for } x \in [ 2 , 2^{k } ] , \\ 1 , & \text{for } x \geq 2^{k}. \end{cases } \end{split}\ ] ] in the following @xmath143 are i.i.d . random variables with distribution function @xmath144 , and @xmath145 stands for their partial sums . for the moments we have ( see ( 29 ) in @xcite ) @xmath146 introduce the infinitely divisible random variables @xmath147 , @xmath148 $ ] , with characteristic function @xmath149,\ ] ] with @xmath150 and @xmath151 . according to corollary 2 in @xcite @xmath147 s are the possible subsequential limits of @xmath152 , more precisely @xmath153 we show an exponential tail bound for the sums conditioned on the maxima . [ lemma : sntail ] for @xmath154 , put @xmath155 for any @xmath64 , @xmath156 and @xmath157 , we have that @xmath158 where @xmath159 for any @xmath160 , we apply the chernoff bounding technique : @xmath161 \\ & = e^{-\lambda \left(x+\e x^{(\lceil \log_2 n \rceil + j)}\right ) } \left ( \e \exp \left [ \frac{\lambda}{n } x^{(\lceil \log_2 n \rceil + j ) } \right ] \right)^{n}. \end{split}\ ] ] one has that @xmath162 & = 1+\frac{\lambda}{n}\e x^{(\lceil \log_2 n \rceil + j ) } + \sum_{\ell=2}^{\infty } \frac { \lambda^{\ell}\e \left\{(x^{({\lceil \log_2 n \rceil + j})})^{\ell } \right\}}{n^\ell \ , \ell ! } \\ & \leq 1 + \frac{\lambda}{n}\e x^{(\lceil \log_2 n \rceil + j ) } + 2 \sum_{\ell=2}^{\infty } \frac { \lambda^{\ell}\left ( \frac{2^{j}}{\gamma_n}\right)^{\ell-1 } } { n\ell ! } \\ & = 1 + \frac{\lambda}{n}\e x^{(\lceil \log_2 n \rceil + j ) } + \frac 2n \frac { e^{\lambda\eta_{j , \gamma_n}}-1-\lambda\eta_{j , \gamma_n}}{\eta_{j , \gamma_n}}\\ & \le \exp \left [ \frac{\lambda}{n}\e x^{(\lceil \log_2 n \rceil + j ) } + \frac 2n \frac { e^{\lambda\eta_{j , \gamma_n}}-1-\lambda\eta_{j , \gamma_n}}{\eta_{j , \gamma_n}}\right],\end{aligned}\ ] ] where we used that by ( [ eq : trunc - moment ] ) @xmath163 \le 2\left ( \frac{n 2^j}{\gamma_n } \right)^{\hspace{-3pt } \ell -1 } \hspace{-4pt},\ ] ] @xmath164 . therefore @xmath165.\end{aligned}\ ] ] with the choice @xmath166 / \eta_{j , \gamma_n}$ ] the lemma is proved . note that @xmath167 , as @xmath1 , therefore the upper bound for large @xmath14 is approximately @xmath168 $ ] . applying the elementary inequalities @xmath169 one has that @xmath170 and so for any @xmath79 @xmath171 since @xmath172 as @xmath173 , for small @xmath157 , we have @xmath174 . we note that this exponential inequality ( and its straightforward extension to generalized st . petersburg games ) allows us to show that arbitrary powers of the random variables @xmath175 are uniformly integrable , whenever @xmath176 . the latter implies that in propositions 2 and 3 in @xcite not only distributional convergence , but also moment convergence holds . going back to ( [ eq : wjg - def ] ) note that each @xmath147 has finite exponential moment of any order . we pointed out in @xcite that the distribution function @xmath177 is infinitely many times differentiable . since the support of the lvy measure is bounded , according to theorem 26.1 in @xcite for the tail behavior of @xmath147 we have the following . for any @xmath178 @xmath179 and so @xmath180 while for @xmath181 @xmath182 and @xmath183 this result combining with proposition 5 in @xcite implies that the tail bound in lemma [ lemma : sntail ] is optimal . expanding the exponential in taylor - series and changing the order of the summation we obtain @xmath184 with @xmath185 as in ( [ eq : eta - def ] ) . the distribution of @xmath147 depends only on the single parameter @xmath186 . denote @xmath187 a random variable with the characteristic function @xmath188 . then , by the definition of @xmath189 @xmath190 thus from the properties of @xmath191 we can derive the properties of @xmath192 , for any @xmath193 . for example , for the density function @xmath194 of @xmath192 we have @xmath195 thus , @xmath194 can be derived from the characteristic function @xmath196 of @xmath197 by a simple transformation . it also shows , that the proper scaling is the variance instead of the standard deviation . from ( [ eq : f ] ) it is apparent that @xmath198 while @xmath199 these limit theorems are in complete accordance with proposition 3 in @xcite , which states that conditioning on small maximum the limit is normal , and with proposition 7 @xcite , which states that conditioning on large maximum the limit is deterministic . by proposition 6 in @xcite for each @xmath41 @xmath200 where @xmath201 with @xmath202 the distribution @xmath203 is poisson - distribution conditioned on being nonzero . from proposition ii.2.7 in @xcite it follows that this distribution is _ not _ infinitely divisible . in theorem 1 in @xcite we showed that for any @xmath204 $ ] @xmath205 for the trimmed sum we have the following the merging theorem , together with the infinite series representation of the limiting distribution function . [ th : trimmed ] we have @xmath206 where @xmath207.\ ] ] since ( [ eq : typical - max - cond ] ) holds uniformly in @xmath14 , we obtain @xmath208 using ( [ eq : max - pr - merge ] ) and the same conditioning as in the proof of theorem 1 in @xcite we obtain the statement . this result implies that , as usual in this setup , along subsequences there is distributional convergence . for the subsequence @xmath209 , @xmath204 $ ] , which in fact covers all the possible limits , this was shown by gut and martin - lf in theorem 6.1 @xcite . the infinite series representation of @xmath29 in theorem 1 in @xcite is in fact equivalent to the distributional representation @xmath210 where @xmath211 , @xmath212 and @xmath213 are independent random variables , @xmath213 has probability distribution @xmath214 , @xmath215 has poisson(@xmath216 ) distribution , conditioned on not being 0 , and @xmath147 is an infinitely divisible distribution given in ( [ eq : wjg - def ] ) . let @xmath217 be a random variable with distribution function @xmath218 . then , the same way ( [ eq : g - star - df ] ) reads as @xmath219 looking at the infinitely divisible random variable @xmath30 as a semistable lvy process at time 1 , the meaning of the representation above is the following . the value @xmath220 corresponds to the maximum jump , @xmath221 is the number of the maximum jumps , and @xmath222 has the law of the lvy process conditioned on that the maximum jump is strictly less than @xmath220 . this kind of distributional representations for general lvy processes were obtained by buchmann , fan and maller , see theorem 2.1 in @xcite . let @xmath223 be i.i.d . exp(1 ) random variables and @xmath224 . [ lemma : ymoment ] for any @xmath225 , the sum @xmath226 converges absolutely with probability 1 and its sum belongs to @xmath227 for any @xmath228 . we have @xmath229 by the hlder inequality we have for any @xmath230 and any @xmath231 with @xmath232 , @xmath233 by the rosenthal inequality ( ( * ? ? ? * theorem 2.9 ) ) we have @xmath234 with some constant @xmath235 depending only on @xmath236 . in the following @xmath237 are positive constants , whose values are not important . on the other hand , @xmath238 is @xmath239 distributed and thus for any @xmath240 we have @xmath241 for @xmath242 , and thus in ( [ 3 ] ) we have @xmath243 since @xmath244 , we can choose @xmath245 so close to 1 that for @xmath246 we have @xmath247 and thus ( [ 5 ] ) holds . choosing @xmath248 close to 1 will make @xmath249 very large , but ( [ 4 ] ) is still valid . therefore , the left hand side of ( [ 3 ] ) is @xmath250 , and consequently in ( [ 2 ] ) we have @xmath251 to estimate @xmath252 we first observe that by large deviation theory we have @xmath253 except on a set @xmath254 with @xmath255 @xmath256 for some absolute constants @xmath257 , @xmath258 . to estimate the difference @xmath259 we have to make sure that @xmath260 and @xmath261 fall into the same dyadic interval . note that when @xmath261 or @xmath260 is close to a discontinuity point of @xmath262 , i.e. to an integer power of 2 , then we can not give a good estimate . therefore assume that @xmath263 . then on the set @xmath264 we have @xmath265 $ ] and thus @xmath266 . therefore , for such @xmath88 s in ( [ 2 ] ) we have @xmath267 except on @xmath254 , and on @xmath254 trivially @xmath268 . thus we proved @xmath269 with @xmath270.\ ] ] for @xmath271 we only have that @xmath272 , but since there are not so many such @xmath88 s it is enough , more precisely @xmath273 consequently by ( [ 2 ] ) , ( [ 6 ] ) we proved that @xmath274 completing the proof of the lemma . in the following theorem we need that the convergence of @xmath275 to some limit @xmath276 is fast enough . however , the natural subsequence @xmath209 satisfies this condition . [ th : trimlim ] assume that @xmath277 along the subsequence @xmath278 , where @xmath31 $ ] . then for any @xmath22 @xmath279 with centering sequence @xmath280 we rewrite the representation ( [ eq : quantile - uni ] ) in terms of the poisson process determined by @xmath281 . since for @xmath2 fix @xmath282 we obtain @xmath283 by the strong law of large numbers @xmath284 a.s . whence it follows @xmath285 now if along a subsequence @xmath286 $ ] we obtain , using ( [ leftside ] ) , @xmath287 note that although @xmath262 is not continuous , the probability that @xmath288 falls in @xmath289 is zero . similar formulas apply for @xmath290 , and thus we get for any fixed @xmath291 @xmath292 observe that @xmath293 now by ( [ a1 ] ) @xmath294 by the strong law of large numbers , the sum in @xmath295 is @xmath296 a.s . , further chebyshev s inequality implies @xmath297 , and thus @xmath298 in probability . rewrite @xmath299 as @xmath300 the second sum obviously converges almost surely to @xmath301 . we show that the first sum converges a.s . to 0 . using @xmath302 , the markov inequality and the borel cantelli lemma , it follows that @xmath303 a.s . here , and in the sequel , constants in the @xmath304 depend only on @xmath305 . thus for any @xmath306 , by the logarithmic periodicity of @xmath262 @xmath307 to estimate @xmath308 we have to use the same idea as in the proof of lemma [ lemma : ymoment ] . given @xmath309 , let @xmath310 be defined by @xmath311 $ ] . then @xmath312 a.s . implies @xmath313 $ ] , moreover @xmath314 $ ] , for @xmath315 . the slope of @xmath316 on this interval is @xmath317 and thus replacing @xmath318 by 1 changes the last expression of ( [ a2 ] ) at most by @xmath319 i.e. , for such @xmath320 s @xmath321 $ ] . here we used that @xmath277 . we emphasize that ( [ a3 ] ) is not true for all @xmath322 . for the sum of these terms @xmath323 the remaining indices can be estimated exactly as in the proof of lemma [ lemma : ymoment ] , and thus we obtain that the first sum in ( [ eq : vnk ] ) converges a.s . on the other hand , for each @xmath320 @xmath324 so the almost sure limit is 0 . the convergence of the series ( [ eq : defyr ] ) imply that @xmath325 a.s . as @xmath326 , completing the proof of the theorem . the definition of the centering sequence @xmath327 is quite natural in view of lemma [ lemma : ymoment ] . however , its asymptotic behavior is not immediately clear , and more importantly , it is not continuous as a function of @xmath276 . this is the reason why we can not prove the merging counterpart of theorem [ th : trimlim ] . in this section we gather some information about @xmath327 . using that @xmath328 we obtain @xmath329 with @xmath330 . thus @xmath331 for the subsequence @xmath332 and for @xmath333 we see that @xmath334 and comparing ( [ eq : sum - merge ] ) and theorem [ th : trimlim ] , this means that @xmath335 . in general , for any @xmath31 $ ] it is easy to see that @xmath336 on the subsequence @xmath209 , that is @xmath337 combined with ( [ eq : an - form ] ) this implies that our centering is ( apart from a constant ) the same as the usual @xmath338 centering in ( [ eq : sum - merge ] ) . more precisely , for any @xmath31 $ ] , @xmath209 , @xmath339 in the limit in ( [ eq : anlimit ] ) appears the function @xmath340.\ ] ] the function @xmath341,\ ] ] was introduced by csrg and simons @xcite , see also kern and wedrich @xcite . here @xmath342 , where @xmath343 s are the dyadic digits of @xmath276 . this function appears naturally if one considers steinhaus resolution of the st . petersburg paradox , see @xcite . csrg and simons ( * theorem 3.1 ) show that @xmath344 is right - continuous , left - continuous except at dyadic rationals greater than @xmath345 and has unbounded variation . moreover , kern and wedrich ( * ? ? ? * theorem 3.1 ) proved that the hausdorff and box - dimension of the graph of @xmath344 is 1 , meaning that @xmath344 is not so irregular . these properties remain true for our function @xmath346 , as it turns out that @xmath347 for any @xmath342 we consider the representation , which contains infinitely many 1 s . then @xmath348 and so @xmath349 thus @xmath350 the limit appearing in ( [ eq : anlimit ] ) is exactly @xmath351 . in this section we obtain a uniform tail bound for the centralized and normalized trimmed sum . [ th : trimtail ] for any @xmath352 , @xmath353 and @xmath64 there is a finite constant @xmath354 , such that @xmath355^{-(r+1 ) } } + c \ , \delta^{-(r+3/2 ) } x^{-(r+3/2)}.\ ] ] as before , consider the representation @xmath356 the tail of the first term is of order @xmath357 ( uniformly in @xmath2 ) . we show that the @xmath227 norm of the remaining sum is bounded , that is for @xmath358 @xmath359 we use the same technique as in the proof of lemma [ lemma : ymoment ] , the only difference is that we need a uniform bound for the empirical quantile process . write @xmath360 for @xmath361 with @xmath362 @xmath363 for the last factor we have @xmath364 for the first factor in ( [ eq : uni - bound ] ) we use mason s inequality ( ( * ? ? ? * proposition 2 ) ) with @xmath365 , @xmath366 , and we get @xmath367 since changing @xmath368 to @xmath2 makes an error @xmath369 , we obtain @xmath370 which is summable . the term @xmath252 can be handled the same way as in the proof of lemma [ lemma : ymoment ] , and we obtain ( [ eq : snr - unibound ] ) . putting @xmath371 , more precise calculation gives @xmath372 thus , using markov s inequality combined with ( [ eq : snr - unibound ] ) , we obtain for any @xmath89 @xmath373^{-(r+1 ) } } + c_{r+3/2 } \ , ( x\delta)^{-(r+3/2 ) } , \end{split}\ ] ] and the statement is proved . introduce the notation @xmath374 using lemma [ lemma : ymoment ] we can determine the tail distribution of the trimmed limit . the tail behavior of the semistable limit along the subsequence @xmath375 ( @xmath376 fix , @xmath131 ) was determined by martin - lf ( * ? ? ? * theorem 4 ) . our proof in the general @xmath4-trimmed setup and also the proof of theorem [ th : snr - tail ] use the same idea as martin - lf : conditioning on the maximum term . [ th : gtail ] for any @xmath377 @xmath378 . \end{split}\ ] ] simple calculation shows that for any @xmath379 @xmath380 and by lemma [ lemma : ymoment ] @xmath381 we have for @xmath14 large enough @xmath382 where we introduced the notation @xmath383 , for @xmath384 . since @xmath385 , by ( [ eq : i1est ] ) @xmath386 conditioning on @xmath387 we have @xmath388 therefore , for @xmath389 @xmath390 by passing to the limit we used the absolute continuity of @xmath391 . finally , @xmath392 martin - lf ( lemma 1 in @xcite ) showed that the left tail is exponentially small ( see also theorem 5 by watanabe and yamamuro @xcite in general ) , thus for second term we have @xmath393 therefore @xmath394 combining the asymptotics the theorem follows . even in the untrimmed case our theorem refines the results ( for general semistable laws ) by watanabe and yamamuro @xcite . for @xmath136 according to theorem [ th : gtail ] we have @xmath395.\ ] ] by ( [ eq : levy - func ] ) we see that exactly the tail of the lvy measure appears , and we have @xmath396 from this we easily obtain @xmath397 which is the statement of theorem 2 in @xcite . from ( [ eq : r0tail ] ) also follows that @xmath398 which is theorem 3 ( i ) in @xcite . however , note that we determine the exact asymptotics of the ratio , and not only the limsup and liminf of it . if @xmath399 for some @xmath89 , then @xmath400 , and so the term corresponding to @xmath401 in ( [ eq : grtail - as ] ) converges to 0 . while if @xmath402 for some @xmath89 , then @xmath403 , and so the term corresponding to @xmath404 in ( [ eq : grtail - as ] ) converges to 1 thus the asymptotic has a simple form when @xmath405 is not close to a power of 2 . in particular for any @xmath406 we have @xmath407 moreover , for @xmath408 @xmath409 thus ( [ eq : grtail - as ] ) reads as @xmath410,\ ] ] as @xmath131 . in the untrimmed case ( @xmath136 ) for @xmath411 this gives @xmath412 , \quad \text{as } m \to \infty,\ ] ] which is exactly martin - lf s asymptotics ( * ? ? ? * theorem 4 , formula ( 9 ) ) . the idea of our proof of theorem [ th : trimlim ] and the representation of the limit goes back to lepage , woodroofe and zinn . let @xmath413 be i.i.d . random variables from the domain of attraction of an @xmath414-stable law , @xmath415 . that is @xmath416 with @xmath417 $ ] , @xmath418 . let @xmath46 denote the partial sum , and let @xmath419 and @xmath420 such that @xmath421 converges in distribution to an @xmath414-stable law @xmath422 . let @xmath423 denote the monotone reordering of @xmath424 . lepage , woodroofe and zinn ( * ? ? ? * theorem 1 ) proved that the limit has the representation @xmath425 where @xmath426 are i.i.d . @xmath427 random variables with @xmath428 , and independently of @xmath429 s @xmath430 are i.i.d . exp(1 ) random variables and @xmath431 . moreover , @xmath432 in case of the two - sided ( symmetric ) version of the st . petersburg game similar results were obtained by berkes , horvth and schauer ( * ? ? ? * corollary 1.4 ) . in this last section we consider some of the previous results in a more general setup , in the case of the so - called generalized st . petersburg game . since the proofs are similar to the proofs in the classical case , we omit them . in this setup peter tosses a possibly biased coin , where the probability of heads at each throw is @xmath433 , and paul s winning is @xmath434 , if the first heads appears on the @xmath6 toss , where @xmath7 , and @xmath435 is a payoff parameter . the classical st . petersburg game corresponds to @xmath436 and @xmath437 . if @xmath8 denotes paul s winning in this st . petersburg@xmath438 game , then @xmath439 , @xmath10 . in this section @xmath18 are i.i.d . petersburg@xmath438 random variables , and @xmath46 , @xmath47 , and @xmath3 stands for the partial sum , partial maximum , and the @xmath4-trimmed sum , respectively . for general @xmath441 we do not have a closed formula for the probabilities @xmath123 . nevertheless , it turns out that the generalized st . petersburg distributions are not subexponential for any choice of the parameters . goldie , c. m. subexponential distributions and dominated - variation tails . _ journal of applied probability _ , * 15 * , no . 2 , pp . 440442 , 1978 . goldie , c. m. , and klppelberg , c. subexponential distributions . in : _ a practical guide to heavy tails_. pp . 435459 , birkhuser boston inc . cambridge , ma , usa , 1998 .	we provide exact asymptotics for the tail probabilities @xmath0 as @xmath1 , for fix @xmath2 , where @xmath3 is the @xmath4-trimmed partial sum of i.i.d . st . petersburg random variables . in particular , we prove that although the st . petersburg distribution is only o - subexponential , the subexponential property almost holds . we also determine the exact tail behavior of the @xmath4-trimmed limits . _ keywords : _ st . petersburg sum ; trimmed sum ; tail asymptotic ; semistable law + _ msc 60f05 , msc 60e07 _
here we show that under quite generic circumstances there is always a sub - gap state bound to magnetic impurities . bound states in a lattice are obtained by solving for the poles of the t - matrix or the zeros of @xmath144 where in the case of a single - lattice site impurity , @xmath145 is calculated on one lattice site . in this case , the single - site green function can be expanded in terms of bdg eigenstates of the bulk as @xmath146 where @xmath147 is the dos of bdg quasiparticles and @xmath148 . following ref . , we note that in the presence of rotational symmetry about the @xmath65-direction the green function @xmath149=0 $ ] , so that it commutes with the impurity . in this case the two operators in the @xmath69-matrix can be simultaneously diagonalized and a bound state occurs whenever @xmath150 , for some eigenvalue @xmath151 of @xmath145 . since we are interested in weak impurities @xmath152 , which could produce states at best at the gap edge , we ignore the energy dependence of @xmath153 and notes that the eigenvalues @xmath154 , where @xmath155 is divergent and @xmath156 are eigenvalues of @xmath157 . since @xmath157 has at least one finite ( non - zero ) eigenvalue , @xmath158 ( as @xmath159 ) where @xmath160 is the gap edge . this implies that there is a bound state at arbitrarily small impurity strengths insde the spectral gap independent of how far the gap is suppressed . near the gap edge , where in 2d , @xmath147 generically has a step function behavior at a minimum and a divergence in our ( i.e. rotationally symmetric bcs ) case one can approximate the energy dependence of @xmath161 .	we show that robust zero - energy bound states can be induced at magnetic impurities in iron - based superconductors with unconventional ( sign - changed ) @xmath0-wave pairing symmetry in the presence of spin - orbit coupling . the zero - energy bound states localized at the impurities are robust to applied magnetic fields and variation in the strength of the impurity potentials by the combined effects of spin - orbit coupling and the existence of low - energy sub - gap states inside the superconducting gap . this work provides an explanation to the puzzling recent observation of robust zero bias peaks observed in scanning tunneling microscopy experiments at magnetic impurities in iron - based superconductor fe@xmath1(te , se ) where @xmath2 is the concentration of magnetic impurities . we are motivated by a recent scanning tunneling microscope ( stm ) observation of a robust zero bias conductance peak ( un - split by an applied magnetic field @xmath3 ) induced at magnetic impurities in iron - based superconductor fe@xmath1(te , se ) @xcite . zero or low - energy sub - gap states bound to magnetic and/or non - magnetic impurities in superconductors are not unusual @xcite . however , in spin - singlet superconductors [ e.g. , @xmath0-wave ( @xmath4-wave ) , @xmath5-wave , etc ] , the zero bias peaks localized at impurities are expected to split into a double peak structure on application of a magnetic field . qualitatively , the splitting of the peak by a magnetic field is due to the fundamental two - fold spin degeneracy of bogoliubov quasi - particle states in a singlet superconductor . since the spin of the zero - energy bound states couples to a magnetic field via zeeman coupling , the zero bias stm peak , if any , gives rise to a double peak structure by application of a magnetic field . theoretically , a defect - induced zero - energy state can escape splitting when the state is non - degenerate . a non - degenerate zero energy bound state in a superconductor , on the other hand , is very unusual , and is commonly referred as a majorana bound state ( mbs ) that can be realized in a topological superconductor @xcite . this has led to the tantalizing conjecture of realizing a topological superconductor and mbs in iron - based superconductors induced by superconductivity , spin - orbit coupling ( which may be induced by the heavy @xmath6 te atoms in fe@xmath1(te , se ) ) , and local magnetic order induced by the magnetic impurities @xcite . in this paper we show that the robust zero - energy bound states localized at magnetic impurities in iron - based superconductor fe@xmath1(te , se ) can be explained in terms of non - topological zero energy bound states which are , nonetheless , robust to perturbations such as zeeman fields and variation of the strength of the impurity potential by the combined effects of spin - orbit coupling ( soc ) and a finite density of low energy sub - gap states present at the impurity sites . as noted already in ref . [ ] , unlike fese , the @xmath6 heavy te atoms in fe(te , se ) has strong soc and may induce the same in the conduction band of fe(te , se ) by modifying the band structure . the low - energy sub - gap states at the impurity sites are either a consequence of soc ( for large enough soc , the system is a nodal superconductor ) or may naturally arise from overlap of the impurity - induced sub - gap states bound to nearby magnetic impurities in the superconductor . we note that the local density of states ( ldos ) measured at the impurity sites in samples with a finite density of magnetic impurities indeed indicates a finite density of low - energy sub - gap quasiparticle states , in addition to a robust zero bias peak un - split by a magnetic field @xcite . with this minimal , experimentally motivated , modeling of the band - structure we are able to show that a robust zero bias conductance peak is induced at magnetic impurities in iron - based superconductors provided the symmetry of the superconducting order parameter is sign - changing @xmath7-wave . on the other hand , the conventional @xmath8 pairing symmetry does not provide the robust zero - energy bound states at magnetic impurities , being strongly affected by the perturbation of a magnetic field even with spin - orbit coupling and a finite density of low energy quasi - particle states . thus , our results , in addition to providing a consistent theoretical explanation of the puzzling observation of robust zero bias conductance peaks in iron - based superconductors ( that has been described as a serious theoretical challenge in recent literature @xcite ) , also helps in identifying the relevant symmetry of the superconducting order parameter as sign - changing @xmath7 wave . we start with a mean - field hamiltonian for the iron - based superconductor using the two - orbital model ( @xmath9 and @xmath10 ) on the two - dimensional fe square lattice @xcite , @xmath11 here , @xmath12 is the tight - binding hamiltonian in the superconducting state , including intra- and inter - orbital hopping integrals @xmath13 where @xmath14 creates an electron with spin @xmath15 in the orbitals @xmath16 ( @xmath9 ) and @xmath17 ( @xmath10 ) at site @xmath18 . we choose the nearest - neighbor hopping matrix elements as @xmath19 , @xmath20 , and the next - nearest - neighbor hopping as @xmath21 , and @xmath22 with @xmath23 . we have taken @xmath24 mev as the energy units and lattice constant @xmath25 . the chemical potential @xmath26 is adjusted to give a fixed filling factor @xmath27 per site . in the unfolded bz , and ( b ) the fermi surfaces of the helicity bands in the presence of the rashba - type soc with @xmath28 mev . the red pockets at @xmath29 and @xmath30 points represent the holelike pockets , and the blue pockets at @xmath31 point represent the electronlike pockets . the soc produces the additional fermi surfaces with the same chemical potential . ] the fermi surfaces consist of the hole pockets at @xmath29 and @xmath30 points and the electron pocket at @xmath31 point in the unfolded brillouin zone [ fig . [ fig : fig1](a ) ] with no soc . the pairing hamiltonian @xmath32 is given by @xmath33 where @xmath34 is a mean - field superconducting order parameter . we focus on the unconventional sign - changed @xmath7 pairing symmetry so that @xmath35 @xcite . for a conventional @xmath8 superconductor , @xmath36 . we can rewrite the hamiltonian in the momentum space as @xmath37 , where @xmath38 and @xmath39 is an 8-dimensional nambu spinor @xmath40 $ ] with @xmath41 being a fourier transform of @xmath42 . the pauli matrices @xmath43 , @xmath44 and @xmath45 act on the orbital , the particle - hole and the spin spaces , respectively . then , the order parameter @xmath46 for @xmath7 pairing , and the dispersions @xmath47 , @xmath48 , @xmath49 , and @xmath50 . the second term in the hamiltonian @xmath51 describes the effects of the magnetic field and the soc . in our work , we consider the out - of - plane magnetic field @xmath52 , and the rashba - type soc @xcite with an angular momentum @xmath53 @xcite on the two - dimensional @xmath54 plane . then , we have @xmath55 , where @xmath56 with @xmath57 being the strength of the soc . here , we treat the magnetic impurity as a localized spin in the classical limit ( @xmath58 ) @xcite , and the quantum ( kondo ) effect of impurity is not under our consideration . in this limit , the magnetic impurity is equivalent to the local magnetic moment @xmath59 . then the impurity hamiltonian describes the interaction between the conduction electrons and the impurity spin located at @xmath60 @xmath61 where @xmath62 and @xmath63 are the intra- and inter - orbital exchange couplings , and the operators @xmath64 . spin - rotational symmetry of the system enable us to choose the @xmath65 axis of the spin degrees of freedom to point in the direction of @xmath59 . we shall consider the effects of the intra - orbital impurity scattering , thus the strength of the impurity is given by @xmath66 . using the nambu spinor @xmath39 , the impurity hamiltonian can be rewritten as @xmath67 with @xmath68 . in our work , we perform a numerical study employing a mean - field @xmath69-matrix approximation @xcite . in this case , we assume that the spatial variation of the superconducting order parameter can be neglected . since the impurity interaction is limited to one site , scattering of quasiparticles from the impurity moment is described by a @xmath69 matrix , @xmath70 whose fourier transform is independent of wave vectors . the single - particle green s function for an impurity located at @xmath71 is given by @xmath72 where @xmath73 . the single - particle green s function for a clean system @xmath74^{-1}$ ] , where @xmath75 is a 8-dimensional identity matrix . the fermi surfaces in the presence of the soc at zero magnetic field is presented in fig . [ fig : fig1](b ) . the helicity bands remove the degeneracies of the electron spin in the electronlike and holelike pockets . with @xmath76 in hands , the @xmath69 matrix can be obtained from the lippmann - schwinger equation @xmath77^{-1}. \label{eq : lippmann}\ ] ] note that as far as the spatial variation of order parameter can be neglected , these equations allow a complete solution of the problem . the nature of the magnetic impurity induced bound states can be found by computing the spin - resolved local density of states ( ldos ) @xmath78 of which the poles give the energy spectra of single - particle excitations , and consist of those of the @xmath79 and the @xmath69 matrix . the poles of @xmath69 matrix signifies the appearance of the impurity induced states . it is known that a strong scattering yields localized states deep in the gap , while a weak scattering results in bound states close to the gap edge @xcite . for a quantum spin , one needs to address the kondo effect @xcite . however , here we treat the magnetic impurity with a classical spin in a mean - field @xmath69 matrix approximation approach . in this case the main effect of the exchange coupling between the local moment @xmath80 and electron spin is the renormalization of the effective scattering potential for electrons of two different spin orientations , and so there are four impurity induced in - gap states , one for each electron - spin orientation in each @xmath9 ( @xmath16 ) and @xmath10 ( @xmath81 ) orbitals . it is noteworthy that the degeneracy between two orbitals are not broken by magnetic field and soc as well , thus we omit the notation of @xmath82 in the spin resolved ldos throughout this paper . for @xmath7 pairing vs bias energy @xmath83 for clean system ( green and magenta for @xmath84 , respectively ) and system with an impurity at @xmath85 ( blue and red for @xmath84 ) at @xmath86 k. ( a ) and ( d ) show the appearance of both zbps for @xmath87 and @xmath88 at the critical values @xmath89 and @xmath90 , respectively ; ( b ) and ( c ) illustrate the zeeman splitting of @xmath91 and @xmath92 at weak magnetic fields @xmath93 and @xmath94 with no soc . whereas ( e ) illustrates the robustness of the zbps , @xmath95 , to the applied magnetic field in the presence of soc , @xmath96 mev , ( f ) the zeeman split begins to appear with increasing magnetic field for a single impurity in the system . comparing the ldos for clean systems ( green and magenta ) , it is manifest that soc slightly changes the low energy states by forming a v - shaped ldos , and make the system robust against magnetic field . ] we begin with the effects of the applied magnetic field and the soc on the in - gap bound states induced by a single magnetic impurity for the sign changed @xmath7-wave iron - based superconductor . fig . [ fig : fig2 ] presents the spin - resolved ldos @xmath97 at the critical values of impurity strengths @xmath89 and @xmath90 for a finite temperature @xmath86 k below the superconducting critical temperature @xmath98 . for a weak impurity scattering @xmath99 , the ground state has time - reversed pairs of single - particle in - gap states @xmath100 , and as @xmath101 increases , the energy @xmath102 approaches the chemical potential and , eventually , at the critical scattering strength @xmath103 , it becomes a zero energy state , or zero - energy bound state ( zbs ) @xcite . fig . [ fig : fig2](a ) and ( d ) show the zbss for both spins ( @xmath87 and @xmath104 ) with zero magnetic field applied , @xmath105 . in the absence of the soc ( @xmath106 ) , the corresponding zero bias peaks ( zbps ) in the ldos at @xmath105 begin to split with increasing magnetic field @xmath52 . fig . [ fig : fig2](b ) and ( c ) illustrate the zeeman splitting by the applied magnetic fields @xmath107 ( @xmath108 tesla ) and @xmath109 ( @xmath110 tesla ) , respectively . in contrast , the presence of the soc ( @xmath111 mev ) dramatically reduces the zeeman splitting and makes the zbps robust to the magnetic field [ fig . [ fig : fig2](e ) and ( f ) ] . it is noteworthy that the zero energy bound states localized at the impurity site in the presence of soc remain pinned to zero energy even in a magnetic field @xmath112 tesla . it is in good agreement with the experimental observations @xcite . we believe that the robust zbss is a strong signature of the presence of the soc in the system . note , however , the low energy quasiparticle states inside the gap in the clean system ( green and magenta ) even at @xmath105 [ fig . [ fig : fig2](d),(e),(f ) ] . they exist because with a finite soc @xmath113 mev , the system is a nodal superconductor . below we will show that in a system with a finite concentration of magnetic impurities as in the experiments @xcite , the low energy quasiparticle density can naturally arise from shiba states bound to the nearby impurities . in this case , the zbss localized at impurity sites are robust to applied magnetic fields ( as well as to variations in the impurity potential ) even for smaller values of soc ( @xmath114 mev ) , corresponding to which the clean system has a full gap . this demonstrates that our results are robust and do not depend on the specific values of the parameters , as long as there is soc and a finite concentration of low energy quasiparticles at the impurity sites as in the experiments . theoretically the robustness of the zbs due to soc can be understood to be due to a combination of suppression of the superconducting gap by soc in the @xmath7 case and the properties of magnetic impurities in nearly isotropic gap superconductors . as seen in fig . [ fig : fig1 ] , soc produces a relatively large spin splitting that pushes one fermi surface towards the x - point and the other towards the @xmath29-point . on the other hand , in @xmath7 pairing , the pairing potential changes sign between these two points in the brillouin zone therefore , as one fermi surface moves towards the @xmath31-point , the pairing amplitude is suppressed , so that one of the fermi surfaces would have a smaller gap . this suppressed gap is still quite isotropic and as shown in the appendix , for such isotropic fermi surfaces , one can prove that an infinitesimal magnetic impurity will support sub - gap states just as in the case of the shiba problem . however , if the gap is suppressed by soc , then the impurity - induced sub - gap state is pinned to live inside the smaller gap , even in the presence of a magnetic field , explaining the robustness of the zero bias peak to zeeman splitting . since the coherence peaks from the larger gapped fermi surfaces are expected to be larger , the smaller gap could appear as a pinned peak inside a larger gap . now we consider the robust zbss induced by the multiple magnetic impurities for @xmath7-wave superconductor in the presence of soc . this addresses the so - called class - d anti - localization mechanism . the case of a large number of bound states to magnetic impurities in a superconductor with soc is described by random - matrix theory in symmetry class d. this symmetry class shows no level repulsion at zero energy . because of this , the disorder averaged density of states ( or the density of states of a large number of weakly localized bound states ) is expected to show a peak at zero energy quite generically independent of the magnetic field . specifically , in the present context , nearby multiple shiba impurities provide a second mechanism ( aside from soc ) of reducing the magnitude of local gap near a given magnetic impurity . by the argument given in the appendix , the zero bias peak at the given impurity will then stay pinned at zero energy even if the value of the soc is smaller , provided there is overlap of wave functions from states localized at nearby impurities . this demonstrates that our results are robust to specific values of soc . below we explore this mechanism numerically by introducing multiple shiba impurities . for @xmath7 pairing vs bias energy @xmath83 for clean system ( green and magenta for @xmath84 , respectively ) and system with multiple impurities ( blue and red for @xmath84 ) at @xmath86 k. 8 impurities are located in a square and @xmath115 atomic sites . the ldos are calculated at a impurity site . ( a)(c ) show the zbps for @xmath87 and @xmath88 at the critical value value of the impurity potential with no soc , @xmath89 . the zbp is split by applying magnetic field @xmath116 ; ( d)(f ) illustrate the robustness of the zbps , @xmath95 in the presence of soc , @xmath117 mev . ] for numerical analysis , we consider 8 impurities arranged in the shape of a square with the nearest neighbors separated by @xmath118 lattice sites ( various other arrangements of impurities give qualitatively the same results ) . for the multiple impurity problem , the green s function , eq . ( [ green_func ] ) , is modified to include the scattering from the neighboring impurities @xmath119 . @xmath120 where @xmath121 runs for all impurities in the system , and @xmath122^{-1}$ ] is the 8@xmath123 matrix with @xmath124 being the number of the neighboring impurities . in fig . [ fig : fig3 ] , we present the spin - resolved ldos @xmath125 at one of the impurity sites ( @xmath119 ) for clean system ( green and magenta lines ) and the impurity induced bound states at the critical scattering @xmath126 ( blue and red lines ) with and without soc . the zeeman splitting in the absence of soc is manifest in the shift of each spin component in the ldos for the clean system with increasing zeeman field , as shown in fig . [ fig : fig3](b ) and ( c ) . as the zeeman field increases , the degeneracy between the two spin components is removed , thus @xmath127 and @xmath128 split from each other . on the other hand , even a weak soc ( @xmath129 ) dramatically changes the zero energy bound states , maintaining the zbps even in the presence of the applied magnetic field @xmath110 tesla . fig . [ fig : fig3](d ) shows the appearance of the zpbs at the critical value of the impurity scattering @xmath130 in the presence of a soc @xmath131 with no magnetic field applied . fig . [ fig : fig3 ] ( e ) and ( f ) illustrate the robustness of the zbps to the magnetic field @xmath52 . pairing at @xmath105 with 8 impurities in a square shape as in fig . [ fig : fig3 ] . ( a ) for @xmath132 , the zbp appears at the critical value of impurity strength @xmath89 . ( b)and ( c ) show the zbp remains in a range of impurity strengths and the robustness increases with increasing soc . ] in addition to robustness to applied magnetic field , in the presence of soc and low energy quasi - particle states within the gap , the zbps in @xmath7-wave superconductors also become robust to variations in the magnitude of the impurity scattering potential . this is important because without the soc the appearance of the zbp at magnetic impurity sites in @xmath7 superconductors requires fine tuned scattering potential at criticality @xmath133 @xcite . in contrast , the soc maintains the appearance of zbp in a range of magnetic impurity scattering potential above the critical value @xmath130 . fig . [ fig : fig4 ] ( b ) and ( c ) show that the presence of the soc maintains the zbps in a broad range of the impurity strength , whereas in the absence of the soc , a zbp is allowed only at the critical impurity scattering @xmath130 [ fig . [ fig : fig4](a ) ] . in this figure , to illustrate the impurity strength dependence of the peak positions , @xmath134 is normalized by the height as @xmath135 . pairing induced by 8 magnetic impurities in the presence of a soc . ( a ) at @xmath126 , the peaks for @xmath87 and @xmath88 are located at zero bias @xmath136 for @xmath105 . ( b ) magnetic field splits into double peaks even in the presence of a soc ( @xmath137 ) . ( c ) and ( d ) illustrate the zbps are sensitive to values of impurity strengths at @xmath137 . insets illustrate the robustness of the zbps for @xmath7 pairing against the magnetic field and impurity strengths at the same soc strength , @xmath138 . ] in contrast to unconventional sign - changed @xmath7 pairing , magnetic impurity - induced zbss in conventional @xmath8 pairing ( @xmath0-wave gap of the same sign on both hole and electron pockets ) is _ not _ robust to applied magnetic fields even in the presence of soc . fig . [ fig : fig5 ] shows the spin resolved ldos @xmath139 at a given impurity site for a spin - orbit coupled superconductor ( @xmath137 ) with @xmath8 pairing ( @xmath140 ) for various values of applied magnetic field and the magnitudes of impurity strengths . fig . [ fig : fig5 ] ( a ) shows the appearance of the zbs at @xmath141 at zero applied magnetic field and ( b ) shows the pronounced zeeman splitting of the peaks even in the presence of soc , which is in stark contrast to the case of @xmath7 pairing as shown in the insets . it is noteworthy that the zbss for the @xmath8 pairing requires fine tuning in the impurity potential even in the presence of soc . fig . [ fig : fig5 ] ( c ) and ( d ) show the double peak structure for @xmath87 and @xmath104 for @xmath142 and @xmath143 , respectively , while the insets show the robustness of the zbss with soc to variations in the magnitude of the impurity potentials for @xmath7 symmetry of the order parameter . in summary , we have shown that robust zero energy bound states are induced at magnetic impurity sites in unconventional sign - changed @xmath7 superconductors in the presence of spin - orbit coupling . the robustness to magnetic fields and variations in the impurity scattering potentials are robust consequences of spin - orbit coupling along with low energy quasiparticle states within the superconducting gap , and do not require the existence of majorana bound states on the surface of iron based superconductors . it is important to note that the zero energy bound states localized at magnetic impurity sites in conventional sign - unchanged ( @xmath8 ) superconductors remain sensitive to applied magnetic fields and variations in the impurity potentials even in the presence of soc . our results , in addition to providing a theoretical explanation of the puzzling observation of robust zero bias peaks in iron based superconductors described earlier as a serious theoretical challenge @xcite , helps identifying the order parameter symmetry of these superconductors as sign - changing @xmath7 wave . k.s and s.t are supported by afosr ( fa9550 - 13 - 1 - 0045 ) . j.d.s would like to acknowledge the university of maryland , condensed matter theory center , and the joint quantum institute for startup support . 10 j - x . yin , _ et . al . _ , nature physics , * 11 * , 543 ( 2015 ) . a.v . balatsky , m.i . salkola , and a. rosengren , phys . rev . b , * 51 * , 15547 ( 1995 ) . a. yazdani , b. a. jones , c. p. lutz , m. f. crommie , and d. m. eigler , science * 275 * , 1767 ( 1997 ) . a. v. balatsky , i. vekhter , and jian - xin zhu , rev . mod . phys . , * 78 * , 373 ( 2006 ) . wei - feng tsai , yan - yang zhang , chen fang , and jiangping hu , phys . rev . b , * 80 * , 064513 ( 2009 ) . y. bang , h .- y . choi , and h. won , phys . rev . b * 79 * , 054529 ( 2009 ) . j. li and y. wang , europhys . lett . * 88 * , 17009 ( 2009 ) . a. akbari , i. eremin , and p. thalmeier , phys . rev . b * 81 * , 014524 ( 2010 ) . t. k. ng and y. avishai , phys . rev . b * 80 * , 104504 ( 2009 ) . m. matsumoto , m. koga , and h. kusunose , j. phys . . jpn . * 78 * , 084718 ( 2009 ) . d. zhang , phys . rev . lett . * 103 * , 186402 ( 2009 ) . t. kariyado and m. ogata , j. phys . . jpn . * 79 * , 083704 ( 2010 ) . r. beaird , i. vekhter , and j .- x . zhu , phys . rev . b * 86 * , 140507(r ) ( 2012 ) . n. read and d. green , phys . rev . b * 61 * , 10267 ( 2000 ) a.y . kitaev , physics - uspekhi * 44 * , 131 ( 2001 ) . e. majorana , nuovo cimento * 14 * , 171 ( 1937 ) . s. raghu , x .- l . qi , c .- x . liu , d. scalapino , and s .- c . zhang , phys . rev . b * 77 * , 220503 ( 2008 ) . k. seo , b. a. bernevig , and j. hu , phys . rev . lett . * 101 * , 206404 ( 2008 ) . y. a. bychkov and e. i. rashba , j. phys . c * 17 * , 6029 ( 1984 ) . l. p. gorkov and e. i. rashba , phys . rev . lett . * 87 * , 037004 ( 2001 ) . a. frigeri , d. f. agterberg , a. koga , and m. sigrist , phys . rev . lett . * 92 * , 097001 ( 2004 ) . l. yu , acta physica sinica * 21 * , 75 ( 1965 ) . h. shiba , progress of theoretical physics * 40 * , 435 ( 1968 ) . a. i. rusinov , soviet journal of experimental and theoretical physics letters * 9 * , 85 ( 1969 ) . jay d. sau and eugene demler , phys . rev . b * 88 * , 205402 ( 2013 ) . f. pientka , l. i. glazman , and f. von oppen , phys . rev . b * 88 * , 155420 ( 2013 ) . polkovnikov , a. , s. sachdev , and m. vojta , phys . rev . lett . * 86 * , 296 ( 2001 ) . zhang , g .- m . , h. hu , and l. yu , phys . rev . lett . * 86 * , 704 ( 2001 ) . zhu , j .- x . , and c. s. ting , phys . rev . b * 63 * , 020506 ( 2001 ) . p. m. r. brydon , s. das sarma , hoi - yin hui , and jay d. sau , phys . rev . b * 91 * , 064505 ( 2015 )
small quantum systems , whose properties are profoundly affected by environment , i.e. , continuum of scattering and decay channels , are intensely studied in various fields of physics ( nuclear physics , atomic and molecular physics , nanoscience , quantum optics , etc . ) . these different open quantum systems ( oqs ) , in spite of their specific features , have generic properties , which are common to all weakly bound / unbound systems close to the threshold . an essential part of the motion of short - lived nucleonic matter is in classically forbidden regions , and their properties are impacted by both the continuum and many - body correlations @xcite . resonances are commonly found in various quantum systems , independently of their building blocks . resonances are genuine intrinsic properties of quantum systems describing preferential decays of unbound states . the effect of resonances and the non - resonant scattering states can be considered in the oqs extension of the shell model ( sm ) , the so - called continuum shell model ( csm ) @xcite . two realizations of the csm have been studied recently : the real - energy csm @xcite and the complex - energy csm @xcite based on the berggren ensemble @xcite , the so - called gamow shell model ( gsm ) . for hermitian hamilton operators , both real - energy csm @xcite and complex - energy csm ( gsm ) @xcite lead naturally to the non - hermitian ( complex - symmetric ) eigenvalue problem . as a result , oqss exhibit several unintuitive properties , which make them qualitatively different from closed quantum systems ( cqs ) . among those salient features are : the segregation of time scales in the continuum @xcite ( see also ref . @xcite for a recent review ) , the alignment of near - threshold states with decay channels @xcite , the instability of sm eigenstates at the channel threshold @xcite , or the resonance crossings @xcite . in this lecture , we shall concentrate on the latter phenomenon which will be illustrated on the example of a non - integrable pairing model . most studies of degeneracies associated with avoided crossings in quantal spectra , focused on the topological structure of the hilbert space and the geometric phases @xcite . among these degeneracies , one finds a diabolic point ( dp ) in hermitian hamiltonians @xcite , and an exceptional point ( ep ) @xcite , which appears in the complex @xmath0-plane of a generic hamiltonian @xmath1 , where both @xmath2 and @xmath3 are hermitian and @xmath4 \neq 0 $ ] . below , we shall introduce a prototypical oqs , the 3-level richardson - gaudin ( rg ) model , to discuss the appearance of resonance crossings and their evolution with a parameter of a non - integrable perturbation . in particular , we shall show the appearance of a new kind of degeneracy , a pseudo - diabolic point ( pseudo - dp ) in the _ anti - hermitian _ limit of this model . rg models @xcite are based on the @xmath5 algebra with elements @xmath6 , @xmath7 , and @xmath8 , fulfilling the commutation relations : @xmath9=\delta_{ll^{\prime } } k^0_l~$ ] , @xmath10=\pm\delta_{ll^{\prime } } k^{\pm}_l~$ ] , where indices @xmath11 refer to a particular copy from a set of @xmath12 , @xmath5 algebras . each @xmath5 algebra possesses one quantum degree of freedom . in the following , we shall use the pair representation of su(2 ) algebra leading to pairing hamiltonians . the elementary operators in this representation are the number operators @xmath13 and the pair operators @xmath14 , @xmath15 , defined as : @xmath16 where @xmath17 is the total angular momentum and @xmath18 is the @xmath19-projection . the state @xmath20 is the time reversal of @xmath21 . the relation between the operators of the pair algebra and the generators of the su(2 ) algebra is : @xmath22 where @xmath23 is the particle degeneracy of level @xmath24 . with this correspondence , one can introduce an integrable 3-level pairing hamiltonian as : @xmath25 below , we shall consider the non - integrable version of the 3-level pairing model @xcite : @xmath26 where @xmath27 is a @xmath28-number . let us solve the non - integrable 3-level pairing model for the case of 2 pairs of fermions in a valence space of degeneracy @xmath41 . the energies of three levels are @xmath42 , @xmath43 , and @xmath44 . in this model space , there are 4 many - body states . fig . [ fig22 ] shows the global pattern of level crossings in the complex @xmath0-plane for @xmath45 . in the lower half - plane of @xmath0 , all eigenvalues are either discrete states at the real @xmath0-axis or decaying resonances . complex conjugate degeneracies situated in the upper half - plane ( @xmath46 ) correspond to capturing resonances . [ fig23 ] shows a typical pattern associated with the formation / breakup of the pseudo - dp at @xmath47 . with an increasing value of @xmath48 , the two eps in the complex @xmath0-plane approach each other , coalesce at @xmath45 and , subsequently , move along the axis @xmath49 for @xmath50 . the coalescence of two eps at @xmath45 , leads to the formation of a double - root singularity , for which the geometric phase associated with a cyclic evolution changes twice faster than for the ep . on the other hand and in contrast to the ordinary dp , both eigenvalues _ and _ eigenvectors merge , like for an ordinary ep . the upper part of fig . [ fig24 ] shows real parts of phase trajectories around the pseudo - dp for all four eigenvectors in the model space . at the pseudo - dp , real part of their eigenenergies are identical . these trajectories in @xmath0-plane are described by : @xmath51 , where @xmath52 defines the way of encircling the degeneracy . the phase of each vector is defined with respect of its value at @xmath53 . the eigenvalues for vectors 2 and 3 are degenerate at the pseudo - dp and change their phases by @xmath54 at each encircling . simultaneously , phases of vectors for non - degenerate eigenvalues 1 and 4 remain approximately constant . for a comparison , the lower part of fig . [ fig24 ] exhibits real parts of phase trajectories of all four vectors around the ep @xmath55 at @xmath56 ( see fig . [ fig23 ] ) . these trajectories are specified by : @xmath57 . eigenvalues for vectors 2 and 3 are degenerate at the ep . after first encircling of the ep they become exchanged . after a second encircling , the phases of vectors 2 and 3 change by @xmath54 , and after two more encircling of the ep all phases return to their initial values . the behavior of eigenenergies in the neighborhood of a pseudo - dp in complex @xmath0-plane are shown along the cut @xmath58 ( see fig . [ fig25 ] ) . the fusion of two eps and the formation of a single pseudo - dp is seen as a sharp crossing of four energies and a coalescence of two widths ( the eigenvectors 2 and 3 ) . even though phases of eigenvectors 1 and 4 are almost constant when encircling the pseudo - dp ( see fig . [ fig24 ] ) , they are essential for the formation of a pseudo - dp degeneracy . it can be shown rigorously , that the pseudo - dp does not result from the jordan block of 2-level system . salient features of eigenvectors around a pseudo - dp of a 3-level pairing hamiltonian ( [ niph ] ) with @xmath45 can be studied analytically . let @xmath59 , where @xmath60 is a complex number . the eigenenergies of a hamiltonian matrix : @xmath61 depend parametrically on the parameter @xmath60 . notice an asymmetric dependence on @xmath60 of eigenvalues 2 and 3 which form a pseudo - dp . the eigenvectors , which are normalized according to the dual metric of rhs @xcite : @xmath62 can be written as : @xmath63 , \nonumber \\ { \hat { { \mbox{\boldmath $ u$}}}}_2 & = & \left [ \begin{matrix } -\frac { ( 1 - \imag ) } { 4\,\sqrt{2\sqrt{2}}}\,\frac{1}{{\sqrt{\delta } } } - 3\,\frac { 1 + \imag } { 4\,\sqrt{\sqrt{2}}}\,{\sqrt{\delta } } + { { \cal o}(\delta)}^{\frac{3}{2 } } \cr \frac{1 + \imag } { 4\,\sqrt{\sqrt{2}}}\frac{1}{{\sqrt{\delta } } } + \frac{1 - \imag } { 2\,\sqrt{2\sqrt{2}}}\,{\sqrt{\delta } } + { { \cal o}(\delta)}^{\frac{3}{2 } } \cr { { \cal o}(\delta)}^{\frac{3}{2 } } \cr \frac{1 - \imag } { 4\,\sqrt{2\sqrt{2}}}\,\frac{1}{{\sqrt{\delta } } } + 3\,\frac{1 + \imag } { 4\,\sqrt{\sqrt{2}}}\,{\sqrt{\delta } } + { { \cal o}(\delta)}^{\frac{3}{2}}\end{matrix } \right ] , \nonumber \\ { \hat { { \mbox{\boldmath $ u$}}}}_3 & = & \left [ \begin{matrix } -\frac{1 + \imag } { 4\,\sqrt{2\,\sqrt{2}}}\,\frac{1}{{\sqrt{\delta } } } - \frac{9\,\sqrt{2 } - 8 - \left ( 9\,\sqrt{2 } + 8 \right ) \,\imag } { 8\,\sqrt{2\,\sqrt{2}}}\,{\sqrt{\delta } } + { { \cal o}(\delta)}^{\frac{3}{2 } } \cr -\frac{1 - \imag } { 4\,2^{1/4}}\,\frac{1}{{\sqrt{\delta } } } + \frac{1 + \imag } { 4\,\sqrt{2\,\sqrt{2 } } } \,{\sqrt{\delta } } + { { \cal o}(\delta)}^{\frac{3}{2 } } \cr \sqrt{3\,\sqrt{2}}\,\left ( 1 + \imag \right ) \,{\sqrt{\delta } } + 11\,\sqrt{\frac{3\,\sqrt{2}}{2}}\,\left ( 1 - \imag \right ) \,{\delta}^{\frac{3}{2 } } + { { \cal o}(\delta)}^{\frac{5}{2 } } \cr \frac{1 + \imag } { 4\,\sqrt{2\,\sqrt{2}}}\,\frac{1}{{\sqrt{\delta } } } + \frac{9\,\sqrt{2 } + 8 - \left ( 9\,\sqrt{2 } - 8 \right ) \,\imag } { 8\,\sqrt{2\,\sqrt{2}}}\,{\sqrt{\delta } } + { { \cal o}(\delta)}^{\frac{3}{2}}\end{matrix } \right ] , \nonumber \\ { \hat { { \mbox{\boldmath $ u$}}}}_4 & = & \left [ \begin{matrix}0.345603 - 0.412057\,\imag - \left ( 2.62262 - 1.08314\,\imag \right ) \,\delta + \left ( 14.2829 + 10.4398\,\imag \right ) \,{\delta}^2 + { { \cal o}(\delta)}^3 \cr -0.582736 + 0.412499\,\imag \,\delta - 6.21696\,{\delta}^2 + { { \cal o}(\delta)}^3 \cr -0.87242 + 3.06004\,\imag \,\delta + 26.4056\,{\delta}^2 + { { \cal o}(\delta)}^3 \cr 0.345603 + 0.412057\,\imag + \left ( 2.62262 + 1.08314\,\imag \right ) \,\delta + \left ( 14.2829 - 10.4398\,\imag \right ) \,{\delta}^2 + { { \cal o}(\delta)}^3\end{matrix } \right ] \nonumber \end{aligned}\ ] ] in the limit @xmath64 , the eigenvectors @xmath65 and @xmath66 with ( identical ) eigenvalues @xmath67 ( see eq . ( [ eigen ] ) ) have divergent components . this feature of eigenvectors at a pseudo - dp makes it similar to an ordinary ep and leads to a singular behavior of various quantities . let us consider for example the pairing operator : @xmath68 , in the neighborhood of a pseudo - dp for @xmath45 : @xmath69 coefficients @xmath70 in this expression are : @xmath71 it is readily seen that all matrix elements of the pairing operator involving states 2 and 3 exhibit the square - root divergence if @xmath64 , i.e. when approaching a pseudo - dp . this divergence is cancelled out if one adds diagonal pairing matrix elements for these two states . however , off - diagonal matrix elements between regular states ( states 1 and 4 in this case ) and any combination of states 2 and 3 are divergent because : @xmath72 and @xmath73 . dependence of a real part of the pairing energy on the interaction strength @xmath0 in the non - integrable pairing hamiltonian ( [ niph ] ) with @xmath45 is plotted along the cut at @xmath58 . pairing energies in strongly mixed eigenstates 2 and 3 have opposite signs and diverge when a pseudo - dp is approached , i.e. if @xmath74 . on the other hand , the sum of pairing energies in states 2 and 3 is both positive and finite and changes smoothly close to a pseudo - dp . these features of pairing energy in the neighborhood of a pseudo - dp are essentially different from those discussed in connection with effects of an ordinary dp on pair - transfer amplitude @xcite as the coupling between eigenvectors involved vanishes at the dp . besides accidental degeneracies , sharp level crossings in the hermitian eigenvalue problem are allowed only between states of different quantum numbers ( different symmetries ) . for the non - hermitian ( complex - symmetric ) eigenvalue problem , such as found in oqss , exact eigenvalue degeneracies appear also for states of the same quantum numbers / symmetries . these are : ( i ) the dp , when riemann sheets on which live eigenvalues just touch each other @xcite , and ( ii ) the ep , when two riemann sheets are entangled by the square - root type of singularity @xcite . we have analyzed here features of degeneracies in the non - hermitian and non - integrable 3-level pairing model ( see eq . ( [ niph ] ) which is a schematic representation of the oqs hamiltonian . in particular , we have shown the appearance of a new kind of degeneracy , the pseudo - dp , in the anti - hermitian limit of this model as a result of the coalescence between two eps . the resonance eigenfunctions at the pseudo - dp ( ep ) are entangled and , except for the ( complex ) eigenenergy , it is impossible to define any physical quantity separately in each of the two resonances involved in the pseudo - dp degeneracy . this entanglement of wave functions may involve many states of an oqs , leading to strong mixing of wave functions for states of largely different eigenenergies . this mechanism is efficient not only in the close neighborhood of a particular pseudo - dp ( ep ) , i.e. for a particular choice of coupling constants in the oqs hamiltonian , but also for systems with largely different coupling constants . future studies , using the csm and realistic effective interactions could provide a first attempt to provide a consistent description of continuum wave functions and related observables in the presence of a pseudo - dp ( ep ) . one of us ( m.p . ) wish to thank witek nazarewicz for useful discussions . this work was supported in part by the spanish mec under grant no . fis2006 - 12783-c03 - 01 and by the cicyt(spanish)-in2p3(french ) cooperation . 99 j. dobaczewski , n. michel , w. nazarewicz , m. poszajczak , and j. rotureau , _ prog . phys . _ * 59 * , 432 ( 2007 ) . j. okoowicz , m. poszajczak , and i. rotter , _ phys . rep . _ * 374 * , 271 ( 2003 ) . k. bennaceur , f. nowacki , j. okoowicz , and m. poszajczak , _ nucl . phys . _ a * 651 * , 289 ( 1999 ) ; _ nucl . phys . _ a * 671 * , 203 ( 2000 ) . j. rotureau , j. okoowicz , and m. poszajczak , _ phys . lett . _ * 95 * , 042503 ( 2005 ) ; _ nucl . phys . _ a * 767 * , 13 ( 2006 ) . t. berggren , _ nucl . phys . _ a * 109 * , 265 ( 1968 ) . r. i d betan , r.j . liotta , n. sandulescu , and t. vertse , _ phys . * 89 * , 042501 ( 2002 ) . n. michel , w. nazarewicz , m. poszajczak , and k. bennaceur , _ phys . lett . _ * 89 * , 042502 ( 2002 ) ; + n. michel , w. nazarewicz , m. poszajczak , and j. okoowicz , _ phys . _ c * 67 * , 054311 ( 2003 ) ; + n. michel , w. nazarewicz , and m. poszajczak , _ phys . _ c * 70 * , 064313 ( 2004 ) . i.m . gelfand and n.ya . vilenkin , _ generalized functions _ , vol . * 4 * , academic press , new york ( 1961 ) . k. maurin , _ generalized eigenfunction expansions and unitary representations of topological groups _ , polish scientific publishers , warsaw ( 1968 ) . a. bohm , _ the rigged hilbert space and quantum mechanics _ , lecture notes in physics * 78 * , springer , new york ( 1978 ) . p. kleinwachter and i. rotter , _ phys . _ c * 32 * , 1742 ( 1985 ) . s. drod , j. okoowicz , m. poszajczak , and i. rotter , _ phys . _ c * 62 * , 4313 ( 2000 ) . r. chatterjee , j. okoowicz , and m. poszajczak , _ nucl . phys . _ a * 764 * , 528 ( 2006 ) . n. michel , w. nazarewicz , and m. poszajczak , _ phys . _ c * 75 * , 031301 ( 2007 ) . yan - an luo , j. okoowicz , m. poszajczak , and n. michel , arxiv : nucl - th/0211068 ; + j. okoowicz , m. poszajczak , and yan - an luo , _ acta phys . pol . _ * 39 * , 389 ( 2008 ) . i. rotter , e. persson , k. pichugin , and p. eba , _ phys . _ e * 62 * , 450 ( 2000 ) . m.v . berry , _ proc . london _ , ser . a * 392 * , 45 ( 1984 ) . lauber , p. weidenhammer , and d. dubbers , _ phys . _ * 72 * , 1004 ( 1994 ) ; d.e . manolopoulos , and m.s . child , _ phys . lett . _ * 82 * , 2223 ( 1999 ) ; + f. pistolesi , and n. manini , _ phys . lett . _ * 85 * , 1585 ( 2000 ) ; + c. dembowski _ et al . _ , * 86 * , 787 ( 2001 ) . berry and m. wilkinson , _ proc . ser . a * 392 * , 15 ( 1984 ) . t. kato , _ perturbation theory for linear operators _ ( springer verlag , berlin , 1995 ) . zirnbauer , j.j.m . verbaarschot and h.a . weidenmller , _ nucl . phys . _ a * 411 * , 161 ( 1983 ) . heiss , and w .- h . steeb , _ j. math . phys . _ * 32 * , 3003 ( 1991 ) . j. dukelsky , s. pittel , and g. sierra , rev . * 76 * , 643 ( 2004 ) . j. dukelsky , j. okoowicz , and m. poszajczak , to be published . nikam , p. ring , and l.f . canto , z. phys . a * 324 * , 241 ( 1986 ) ; + r.s . nikam , p. ring , and l.f . canto , phys . b * 185 * , 269 ( 1987 ) ; + c. price , h. esbensen , and s. landowne , phys . b * 197 * , 15 ( 1987 ) ; + j. de boer , c.h . dasso , and g. pollarolo , z. phys . a * 335 * , 199 ( 1990 ) ; + y. sun , p. ring , and r.s . nikam , z. phys . a * 339 * , 51 ( 1991 ) ; + l.f . c * 47 * , 2836 ( 1993 ) ; + k.g . et al . _ , c * 48 * , 1879 ( 1993 ) .	the evolution pattern of exceptional points is studied in a non - integrable limit of the complex - extended 3-level richardson - gaudin model . the appearance of a pseudo - diabolic point from the fusion of two exceptional points is demonstrated in the anti - hermitian limit of the model and studied in some details .
squares and square roots of graphs form a classical and well - studied topic in graph theory , which has also attracted significant attention from the algorithms community . a graph @xmath0 is the _ square _ of a graph @xmath2 if @xmath0 and @xmath2 have the same vertex set , and two vertices are adjacent in @xmath0 if and only if the distance between them is at most @xmath1 in @xmath2 . this situation is denoted by @xmath3 , and @xmath2 is called a _ square root _ of @xmath0 . a square root of a graph need not be unique ; it might even not exist . there are graphs without square roots , graphs with a unique square root , and graphs with several different square roots . characterizing and recognizing graphs with square roots has therefore been an intriguing and important problem both in graph theory and in algorithms for decades . already in 1967 , mukhopadhyay @xcite proved that a graph @xmath0 on vertex set @xmath4 has a square root if and only if @xmath0 contains complete subgraphs @xmath5 , such that each @xmath6 contains @xmath7 , and vertex @xmath8 belongs to @xmath6 if and only if @xmath7 belongs to @xmath9 . unfortunately , this characterization does not yield a polynomial - time algorithm for deciding whether @xmath0 has a square root . let us formally call square root the problem of deciding whether an input graph @xmath0 has a square root . in 1994 it was shown by motwani and sudan @xcite that square root is np - complete . motivated by its computational hardness , special cases of the problem have been studied , where the input graph @xmath0 belongs to a particular graph class . according to these results , square root is polynomial - time solvable on planar graphs @xcite , and more generally , on every non - trivial minor - closed graph class @xcite . polynomial - time algorithms exist also when the input graph @xmath0 belongs to one of the following graph classes : block graphs @xcite , line graphs @xcite , trivially perfect graphs @xcite , threshold graphs @xcite , graphs of maximum degree 6 @xcite , graphs of maximum average degree smaller than @xmath10 @xcite , graphs with clique number at most 3 @xcite , and graphs with bounded clique number and no long induced path @xcite . on the negative side , it has been shown that square root is np - complete on chordal graphs @xcite . a number of parameterized complexity results exist for the problem @xcite . more interesting from our perspective , the intractability of the problem has also been attacked by restricting the properties of the square root that we are looking for . in this case , the input graph @xmath0 is arbitrary , and the question is whether @xmath0 has a square root that belongs to some graph class @xmath11 specified in advance . we denote this problem by @xmath12-square root , and this is exactly the problem variant that we focus on in this paper . significant advances have been made also in this direction . previous results show that @xmath11-square root is polynomial - time solvable for the following graph classes @xmath11 : trees @xcite , proper interval graphs @xcite , bipartite graphs @xcite , block graphs @xcite , strongly chordal split graphs @xcite , ptolemaic graphs @xcite , 3-sun - free split graphs @xcite , cactus graphs @xcite , and graphs with girth at least @xmath13 for any fixed @xmath14 @xcite . the result for 3-sun - free split graphs has been extended to a number of other subclasses of split graphs in @xcite . observe that if @xmath12-square root is polynomial - time solvable for some class @xmath11 , then this does not automatically imply that @xmath15-square root is polynomial - time solvable for a subclass @xmath16 of @xmath12 . on the negative side , @xmath11-square root remains np - complete for each of the following graph classes @xmath11 : graphs of girth at least 5 @xcite , graphs of girth at least 4 @xcite , split graphs @xcite , and chordal graphs @xcite . all known np - hardness constructions involve dense graphs @xcite , and the square roots that occur in these constructions are dense as well . this , in combination with the listed polynomial - time cases , naturally leads to the question whether @xmath11-square root is polynomial - time solvable if the class @xmath12 is `` sparse '' in some sense . motivated by the above , in this paper we study @xmath12-square root when @xmath12 is the class of outerplanar graphs , and when @xmath12 is the class of graphs of pathwidth at most @xmath1 . in both cases , we show that @xmath12-square root can be solved in polynomial time . in particular , we prove that outerplanar ( square ) root can be solved in time @xmath17 and ( square ) root of pathwidth @xmath18 in time @xmath19 . our approach for outerplanar graphs can in fact be directly applied to every subclass of outerplanar graphs that is closed under edge deletion and that can be expressed in monadic second - order logic , including cactus graphs , for which a polynomial - time algorithm is already known . we consider only finite undirected graphs without loops and multiple edges . we refer to the textbook by diestel @xcite for any undefined graph terminology . let @xmath0 be a graph . we denote the vertex set of @xmath0 by @xmath20 and the edge set by @xmath21 . the subgraph of @xmath0 induced by a subset @xmath22 is denoted by @xmath23 $ ] . the graph @xmath24 is the graph obtained from @xmath0 after removing the vertices of @xmath25 . if @xmath26 , we also write @xmath27 . similarly , we denote the graph obtained from @xmath0 by deleting a set of edges @xmath28 , or a single edge @xmath29 , by @xmath30 and @xmath31 , respectively . the _ distance _ @xmath32 between a pair of vertices @xmath33 and @xmath34 of @xmath0 is the number of edges of a shortest path between them . the _ open neighborhood _ of a vertex @xmath35 is defined as @xmath36 , and its _ closed neighborhood _ is defined as @xmath37 = n_g(u ) \cup \{u\}$ ] . for @xmath38 , @xmath39 . two adjacent vertices @xmath40 are said to be _ true twins _ if @xmath37=n_g[v]$ ] . a vertex @xmath34 is _ simplicial _ if @xmath41 $ ] is a clique , that is , if there is an edge between any two vertices of @xmath41 $ ] . the _ degree _ of a vertex @xmath35 is defined as @xmath42 . the maximum degree of @xmath0 is @xmath43 . a vertex of degree 1 is said to be a _ pendant _ vertex . a _ connected component _ of @xmath0 is a maximal connected subgraph . a vertex @xmath33 is a _ cut vertex _ of a graph @xmath0 with at least two vertices if @xmath27 has more components than @xmath0 . a connected graph without cut vertices is said to be _ biconnected_. an inclusion - maximal induced biconnected subgraph of @xmath0 is called a _ block_. for a positive integer @xmath44 , the _ @xmath44-th power _ of a graph @xmath2 is the graph @xmath45 with vertex set @xmath46 such that every pair of distinct vertices @xmath33 and @xmath34 of @xmath0 are adjacent if and only if @xmath47 . for the particular case @xmath48 , @xmath49 is a _ square _ of @xmath2 , and @xmath2 is a _ square root _ of @xmath0 if @xmath50 . the _ contraction _ of an edge @xmath51 of a graph @xmath0 is the operation that deletes the vertices @xmath33 and @xmath34 and replaces them by a vertex @xmath52 adjacent to @xmath53 . a graph @xmath54 is a contraction of a graph @xmath0 if @xmath54 can be obtained from @xmath0 by a series of edge contraction . a graph @xmath54 is a _ minor _ of @xmath0 if it can be obtained from @xmath0 by vertex deletions , edge deletions and edge contractions . a graph @xmath0 is _ planar _ if it admits an embedding on the plane such that there are no edges crossing ( except in endpoints ) . a planar graph @xmath0 is _ outerplanar _ if it admits a crossing - free embedding on the plane in such a way that all its vertices are on the boundary of the same ( external ) face . for a considered outerplanar graph , we always assume that its embedding on the plane is given . if @xmath0 is a planar biconnected graph different from @xmath55 , then for any of its embeddings , the boundary of each face is a cycle ( see , e.g. , @xcite ) . if @xmath0 is a biconnected outerplanar graph distinct from @xmath55 , then the cycle @xmath56 forming the boundary of the external face is unique and we call it the _ boundary cycle_. by definition , all vertices of @xmath0 are laying on @xmath56 , and every edge is either an edge of @xmath56 or a _ chord _ of @xmath56 , that is , its endpoints are vertices of @xmath56 that are non - adjacent in @xmath56 . clearly , these chords are not intersecting in the embedding . for a vertex @xmath33 , we define the _ canonical ordering with respect to @xmath33 _ as a clockwise ordering of the vertices on @xmath56 starting from @xmath33 . for a subset of vertices @xmath57 , the canonical _ canonical ordering of @xmath57 with respect to @xmath33 _ is the ordering induced by the canonical ordering of the vertices of @xmath56 . see figure [ fig : ord ] a ) for some examples . in our paper we use these terms for blocks of an outerplanar graphs distinct from @xmath55 . outerplanar graphs can also be characterized via forbidden minors as shown by syso @xcite . [ lem : out - minor ] a graph @xmath0 is outerplanar if and only if it does not contain @xmath58 and @xmath59 as minors . a _ tree decomposition _ of a graph @xmath0 is a pair @xmath60 where @xmath61 is a tree and @xmath62 is a collection of subsets ( called _ bags _ ) of @xmath20 such that the following three conditions hold : * @xmath63 , * for each edge @xmath64 , @xmath65 for some @xmath66 , and * for each @xmath67 the set @xmath68 induces a connected subtree of @xmath61 . the _ width _ of a tree decomposition @xmath69 is @xmath70 . the _ treewidth _ @xmath71 of a graph @xmath0 is the minimum width over all tree decompositions of @xmath0 . if @xmath61 is restricted to be a path , then we say that @xmath72 is a _ path decomposition _ of @xmath0 , the _ pathwidth _ @xmath73 of @xmath0 is the minimum width over all path decompositions of @xmath0 . notice that a path decomposition of @xmath0 can be seen as a sequence @xmath74 of bags . we always assume that the bags @xmath74 are distinct and inclusion maximal , that is , there are no bags @xmath75 and @xmath76 such that @xmath77 . the following fundamental results are due to bodlaender @xcite , and bodlaender and kloks @xcite . [ l - bod ] for every fixed constant @xmath78 , it is possible to decide in linear time whether the treewidth of a graph is at most @xmath78 . for every fixed constant @xmath78 , it is possible to decide in linear time whether the pathwidth of a graph is at most @xmath78 . we need the following three folklore observations about treewidth and pathwidth . [ obs : minor ] if @xmath2 is a minor ( contraction ) of @xmath0 , then @xmath79 and @xmath80 . [ obs : tw - op ] for an outerplanar graph @xmath0 , @xmath81 . [ lem : tw - power ] for a graph @xmath0 and a positive integer @xmath44 , @xmath82 and @xmath83 for more information on treewidth and pathwidth we refer to @xcite . let @xmath2 be a square root of a graph @xmath0 . we say that @xmath2 is a _ minimal _ square root of @xmath0 if @xmath84 , and no proper subgraph of @xmath2 is a square root of @xmath0 . we need the following simple observations . [ obs : min ] let @xmath12 be a graph class closed under edge deletion . if a graph @xmath0 has a square root @xmath85 , then @xmath0 has a minimal square root that belongs to @xmath12 . [ obs : triangle ] let @xmath2 be a minimal square root of a graph @xmath0 that contains three pairwise adjacent vertices @xmath86 . then @xmath34 or @xmath52 has a neighbor @xmath87 in @xmath2 such that @xmath88 is not adjacent to @xmath33 . since @xmath2 is a minimal square root of @xmath0 , @xmath89 is not a square root of @xmath0 . because @xmath90 , we conclude that there is @xmath91 such that @xmath2 has a unique @xmath92-path @xmath93 of length two and @xmath94 is an edge of this path . then @xmath95 or @xmath96 for some @xmath87 that is not adjacent to @xmath33 . we conclude this section by a lemma that enables us to identify some edges that are not included in any square root . the lemma is implicit in @xcite ; for the sake of completeness we give a full proof here . [ lem : not - incl ] let @xmath97 be distinct neighbors of a vertex @xmath33 in a graph @xmath0 such that @xmath88 and @xmath98 are at distance at least @xmath99 in @xmath27 . then @xmath100 for any square root @xmath2 of @xmath0 . suppose that @xmath2 is a square root of @xmath0 . to obtain a contradiction , assume that @xmath101 . if @xmath102 , then @xmath103 contradicting the condition that @xmath104 . hence , @xmath105 . because @xmath106 , there are @xmath107 . if @xmath108 , then @xmath103 ; a contradiction . if @xmath109 , then @xmath110 and @xmath111 is a path in @xmath27 of length 2 and again we obtain a contradiction with the condition @xmath104 . in this section we show that it can be decided in polynomial time whether a graph has an outerplanar square root . we say that a square root @xmath2 of @xmath0 is an _ outerplanar root _ if @xmath2 is outerplanar . we define the following problem : outerplanar root : : input : : : a graph @xmath0 . question : : : is there an outerplanar graph @xmath2 such that @xmath84 ? the main result of this section is the following . [ thm : outerplanar ] outerplanar root can be solved in time @xmath17 , where @xmath112 is the number of vertices of the input graph . the remaining part of this section is devoted to the proof of theorem [ thm : outerplanar ] . in section [ sec : tech ] we obtain several structural results we need to construct an algorithm for outerplanar root . then in section [ sec : alg ] we construct a polynomial - time algorithm for outerplanar root . in this section we give several structural results about outerplanar square roots . let @xmath2 be an outerplanar root of a graph @xmath0 , @xmath35 . we say that two distinct vertices @xmath113 are _ consecutive with respect to @xmath33 _ if @xmath88 and @xmath98 are in the same block @xmath114 of @xmath2 and there are no vertices of @xmath115 between @xmath88 and @xmath98 in the canonical ordering of the vertices of the boundary cycle of @xmath114 with respect to @xmath33 . for a set of vertices @xmath116 , we say that the vertices @xmath57 are _ consecutive with respect to @xmath33 _ if the vertices of @xmath57 are in the same block of @xmath2 and any two vertices of @xmath57 consecutive in the canonical ordering of elements of @xmath57 with respect to @xmath33 are consecutive with respect to @xmath33 ; a singleton is assumed to be consecutive ( see figure [ fig : ord ] b for an example ) . as every subgraph of an outerplanar graph is outerplanar , by observation [ obs : min ] , we may restrict ourselves to minimal outerplanar roots . let @xmath2 be a minimal outerplanar root of a graph @xmath0 and let @xmath35 . denote by @xmath117 a collection of all subsets @xmath57 of @xmath115 such that @xmath118 for some @xmath119 . we intend to use the notion @xmath117 to find edges of @xmath0 which can not belong to any square root of @xmath0 . to this aim we study properties of @xmath117 . [ lem : cons ] let @xmath2 be a minimal outerplanar root of a graph @xmath0 , and let @xmath35 . then for each @xmath120 , @xmath57 is consecutive with respect to @xmath33 . let @xmath120 and consider @xmath119 such that @xmath118 . notice that @xmath121 , because @xmath88 should be adjacent to a vertex of @xmath115 since @xmath122 . if @xmath123 , then the claim holds by definition . assume that @xmath124 . notice that the vertices of @xmath57 are in the same block of @xmath2 , because if @xmath125 are not in the same block @xmath114 of @xmath2 , then any vertex adjacent to @xmath98 and @xmath126 in @xmath0 is in @xmath127 $ ] . assume that @xmath128 and the vertices are numbered in the canonical order with respect to @xmath33 . suppose that @xmath57 is not consecutive with respect to @xmath33 . then there is a vertex @xmath98 lying on @xmath56 between @xmath129 and @xmath130 for some @xmath131 such that @xmath102 and @xmath132 . since @xmath133 , @xmath2 has @xmath134 and @xmath135-paths @xmath136 and @xmath137 of length at most @xmath1 . notice that these paths do not contain @xmath33 , because @xmath138 , and they do not contain @xmath98 , because @xmath132 . this implies that @xmath98 is inside of the inner face of @xmath56 in the outerplanar embedding of @xmath0 ; a contradiction . [ lem : non - edges ] let @xmath2 be a minimal outerplanar root of a graph @xmath0 , and let @xmath35 . if for two distinct vertices @xmath113 there is no set @xmath120 such that @xmath139 , then @xmath140 . suppose that for two distinct @xmath113 , @xmath141 . by observation [ obs : triangle ] , there is a vertex @xmath126 such that @xmath126 is adjacent to @xmath88 or @xmath98 in @xmath2 , but @xmath126 is not adjacent to @xmath33 in @xmath2 . we obtain that @xmath88 and @xmath98 are adjacent to @xmath126 in @xmath0 and , therefore , @xmath142 . in other words , @xmath143 . we also need the following two observations . [ lem : adj ] let @xmath2 be a minimal outerplanar root of a graph @xmath0 , and let @xmath35 . if @xmath144 is not a pendant vertex of @xmath2 , then there is @xmath145 that is adjacent to @xmath88 in @xmath0 . since @xmath88 is not a pendant vertex , it has a neighbor in @xmath2 distinct from @xmath33 . if there is a @xmath146 such that @xmath147 $ ] , the claim holds . assume that for every @xmath148 distinct from @xmath33 , it holds that @xmath149 . consider such a neighbor @xmath98 of @xmath33 . by observation [ obs : triangle ] , @xmath98 has a neighbor @xmath126 in @xmath2 such that @xmath150 . since @xmath151 , @xmath110 , we have that @xmath152 . [ lem : size ] let @xmath2 be a minimal outerplanar root of a graph @xmath0 , and let @xmath35 . then any @xmath120 has size at most @xmath153 . suppose that there is a set @xmath154 of size at least 5 . then @xmath118 for @xmath119 . by lemma [ lem : cons ] , @xmath57 is consecutive with respect to @xmath33 , and w.l.o.g . @xmath155 is the canonical order of the vertices of @xmath57 along the boundary cycle @xmath56 of the block of @xmath2 with respect to @xmath33 . suppose that @xmath88 lays on @xmath56 . if it lays before @xmath156 in the canonical ordering of the vertices of @xmath56 with respect to @xmath33 , then @xmath88 is not adjacent to @xmath157 in @xmath0 by outerplanarity ; a contradiction . similarly , if @xmath88 lays after @xmath156 , @xmath88 is not adjacent to @xmath158 in @xmath0 and we also have a contradiction . suppose that @xmath88 does not belong to @xmath56 . then @xmath88 is at distance at most @xmath1 in @xmath2 from @xmath158 and @xmath157 . it follows that there is a @xmath159-path @xmath93 in @xmath2 of length at most @xmath153 avoiding @xmath33 and at least one vertex of @xmath57 but this contradicts outerplanarity . by combining lemmas [ lem : cons ] and [ lem : size ] we obtain the following lemma . [ lem : del ] let @xmath2 be a minimal outerplanar root of a graph @xmath0 , and let @xmath35 . then the following holds . * if @xmath113 do not belong to the same block of @xmath2 , then for any @xmath120 , @xmath160 or @xmath161 . * if @xmath114 is a block of @xmath2 containing @xmath33 and vertices @xmath162 ordered in the canonical order with respect to @xmath33 in the boundary cycle of @xmath114 , then for any @xmath120 , @xmath163 or @xmath164 if @xmath165 and @xmath166 . we now state some structural results that help to decide whether an edge incident to a vertex is in an outerplanar root or not . suppose that @xmath33 and @xmath34 are pendant vertices of a square root @xmath2 of @xmath0 and that @xmath33 and @xmath34 are adjacent to the same vertex of @xmath167 . then , in @xmath0 , @xmath33 and @xmath34 are simplicial vertices and true twins . we use this observation in the proof of the following lemma that allows to find some pendant vertices . [ lem : pend ] let @xmath2 be a minimal outerplanar root of a graph @xmath0 . if @xmath0 contains at least @xmath168 simplicial vertices that are pairwise true twins , then at least one of these vertices is a pendant vertex of @xmath2 . let @xmath2 be a minimal outerplanar root of a graph @xmath0 that contains a set @xmath57 of 7 simplicial vertices that are pairwise true twins . suppose first that @xmath57 contains two vertices @xmath88 and @xmath98 that do not belong to the same block of @xmath2 . we claim that @xmath88 is a pendant vertex of @xmath2 . since @xmath88 and @xmath98 are adjacent in @xmath0 , we obtain that @xmath169 for a cut vertex @xmath33 that belongs to two blocks @xmath170 and @xmath171 of @xmath2 containing @xmath88 and @xmath98 respectively . to obtain a contradiction , assume that @xmath88 has a neighbor @xmath172 in @xmath2 . clearly , @xmath126 is not in @xmath171 . it implies that @xmath173 , because @xmath88 and @xmath98 are true twins of @xmath0 and , therefore , @xmath174 . since @xmath2 is a minimal square root , @xmath175 is not a square root of @xmath0 . because @xmath176 , we have that there is a vertex @xmath177 such that @xmath178 but either i ) @xmath179 , @xmath180 and @xmath181 is the unique @xmath182-path of length at most @xmath1 in @xmath2 or , symmetrically , ii ) @xmath183 , @xmath184 and @xmath185 is the unique @xmath186-path of length at most @xmath1 in @xmath2 . because @xmath88 and @xmath98 are true twins of @xmath0 and @xmath177 , we obtain that @xmath187 and , therefore , @xmath188 . hence , @xmath189 and @xmath190 are paths of @xmath2 contradicting i ) or ii ) respectively . we conclude that @xmath88 is a pendant vertex of @xmath2 . suppose now that the vertices of @xmath57 belong to the same block @xmath114 of @xmath2 . let @xmath56 be the boundary cycle of @xmath114 and denote by @xmath191 the vertices of @xmath57 numbered according to the canonical order with respect to an arbitrary vertex of @xmath56 . because the vertices of @xmath57 are pairwise adjacent in @xmath0 , @xmath114 has a chord @xmath51 such that @xmath57 has vertices in the two components of @xmath192 . among all such chords we choose @xmath51 and a component @xmath193 of @xmath192 in such a way that @xmath193 contains the minimum number of vertices of @xmath57 . let @xmath194 be the other compoent of @xmath192 . assume without loss of generality that @xmath195 and let @xmath196 for @xmath197 be the vertices of @xmath57 in @xmath194 . notice that @xmath194 contains at least 3 vertices of @xmath57 by the choice of @xmath51 . assume also that @xmath34 is after @xmath158 in the canonical ordering of the vertices of @xmath56 with respect to @xmath33 . because the vertices of @xmath57 are adjacent in @xmath0 , they are at distance at most @xmath1 in @xmath2 . it implies that for any @xmath198 and any @xmath199 , @xmath200 or @xmath201 by the outerplanarity of @xmath114 . suppose that @xmath193 contains at least two vertices of @xmath57 . by symmetry , we can assume that @xmath202 . it follows that @xmath203 , @xmath204 but @xmath205 by the choice of @xmath51 . we obtain that @xmath206 and @xmath207 , but these are intersecting chords of @xmath56 ; a contradiction . we conclude that @xmath158 is the unique vertex of @xmath57 in @xmath193 . notice that it implies that @xmath208 and @xmath209 but it can happen that @xmath210 or @xmath211 . we also have that @xmath203 or @xmath212 . using symmetry we assume that @xmath203 . if @xmath213 , then @xmath214 for @xmath215 . hence , @xmath33 is adjacent to @xmath216 in this case . suppose that @xmath217 . we show that either @xmath214 for @xmath215 or @xmath218 for @xmath215 . assume that this is not the case and that there is an @xmath219 such that @xmath220 and @xmath221 such that @xmath222 . since @xmath220 , @xmath223 . because we can not have intersecting chords in @xmath56 , @xmath224 similarly , we obtain that @xmath225 . since , for every @xmath226 , @xmath214 or @xmath218 and @xmath227 , we have that the distance between @xmath130 and @xmath228 in @xmath2 is at least 3 by the outerplanarity of @xmath2 , a contradiction . therefore , the claim holds . by symmetry , we can assume that @xmath214 for @xmath215 . in particular , @xmath33 is adjacent to @xmath216 we now show that @xmath156 is a pendant vertex of @xmath2 . to obtain a contradiction , assume that @xmath156 has a neighbor @xmath98 in @xmath2 distinct from @xmath33 . because @xmath156 and @xmath229 are true twins of @xmath0 , @xmath230 and , therefore , @xmath231 . because @xmath2 is a minimal outerplanar root , @xmath232 is not a square root of @xmath0 . because @xmath233 , it follows that there is a vertex @xmath172 such that @xmath234 but either i ) @xmath235 , @xmath236 and @xmath237 is the unique @xmath238-path in @xmath2 or , symmetrically , ii ) @xmath239 , @xmath240 and @xmath241 is the unique @xmath242-path in @xmath2 . if @xmath235 , then by the same arguments as for @xmath98 , we have that @xmath173 and we get another path @xmath243 . let @xmath240 . if @xmath244 , then @xmath173 and we have the path @xmath245 . hence , @xmath246 . since @xmath126 is adjacent to @xmath156 in @xmath0 , @xmath126 is a neighbor of @xmath229 in @xmath0 . the only possibility is that @xmath247 , i.e. , @xmath126 lays in @xmath56 between @xmath248 and @xmath249 . since @xmath126 is adjacent to @xmath158 , we obtain that @xmath250 . again , we have the path @xmath245 ; a contradiction . we conclude that @xmath156 is a pendant vertex of @xmath2 . we apply lemma [ lem : not - incl ] to identify the edges incident to a vertex of sufficiently high degree in an outerplanar root using the following two lemmas . [ lem : three ] let @xmath0 be a graph having a minimal outerplanar root @xmath2 . let also @xmath35 be such that there are three distinct vertices @xmath251 that are pairwise at distance at least @xmath99 in @xmath27 . then for @xmath252 , @xmath138 if and only if there is an @xmath253 such that @xmath254 . let @xmath2 be a minimal outerplanar root of @xmath0 . if there is an @xmath253 such that @xmath254 , then @xmath138 by lemma [ lem : not - incl ] . hence , to complete the proof , we have to show that if @xmath138 , then there is @xmath253 such that @xmath254 . observe that this property trivially holds if @xmath255 for some @xmath253 . assume that @xmath256 . notice that @xmath257 by lemma [ lem : not - incl ] for @xmath253 . let @xmath252 such that @xmath138 . suppose that there is an @xmath253 such that @xmath88 and @xmath7 are in distinct components of @xmath258 . then any @xmath259-path in @xmath2 goes through @xmath33 and has length at least 4 , because @xmath260 . it follows that @xmath254 . suppose now that @xmath261 and @xmath88 are in the same component of @xmath258 . since @xmath262 , there is @xmath149 such that @xmath141 . let @xmath114 be a block of @xmath2 containing @xmath33 and @xmath98 and denote by @xmath56 its boundary cycle . because @xmath261 and @xmath88 are in the same component of @xmath258 , each @xmath7 is either i ) a vertex of @xmath114 and we let @xmath263 in this case or ii ) @xmath264 and there is a unique @xmath265 such that @xmath266 and @xmath267 ( see figure [ fig : case - one - comp ] ) . assume that @xmath268 is the canonical order with respect to @xmath33 in @xmath56 . let @xmath269 . if @xmath98 lays before @xmath270 on @xmath56 in the canonical ordering with respect to @xmath33 , then @xmath271 . similarly , if @xmath98 lays after @xmath270 on @xmath56 in the canonical ordering with respect to @xmath33 , then @xmath272 . suppose that @xmath273 . notice that it means that @xmath274 . if @xmath275 , then @xmath98 is the unique neighbor of @xmath88 in @xmath2 that is in @xmath114 . we have that @xmath276 . if @xmath277 , then @xmath88 is either before or after @xmath270 in the canonical ordering with respect to @xmath33 . we have that @xmath271 in the first case and @xmath272 in the second . [ lem : high ] let @xmath0 be a graph having a minimal outerplanar root @xmath2 such that any vertex of @xmath2 has at most @xmath168 pendant neighbors . let also @xmath35 with @xmath278 . then there are distinct @xmath251 that are pairwise at distance at least @xmath99 in @xmath27 . let @xmath2 be a minimal outerplanar root of @xmath0 such that any vertex of @xmath2 has at most @xmath168 pendant neighbors and let @xmath35 be a vertex with @xmath278 . denote by @xmath279 the set of pendant neighbors of @xmath33 and let @xmath280 . notice that @xmath281 , because @xmath282 . suppose that the vertices of @xmath57 are in at least 3 distinct components of @xmath258 . then there are distinct blocks @xmath283 , @xmath284 and @xmath285 of @xmath2 containing @xmath33 and at least one vertex of @xmath57 each . by lemma [ lem : adj ] , there are @xmath286 such that @xmath7 is adjacent to a vertex of @xmath287 in @xmath0 for @xmath253 . notice that @xmath288 , @xmath289 and @xmath290 are in distinct components of @xmath258 . we obtain that they are pairwise at distance at least 3 in @xmath27 . suppose now that the vertices of @xmath57 are in exactly two distinct components of @xmath258 . then there are two blocks @xmath283 and @xmath284 of @xmath2 containing @xmath33 and the vertices of @xmath57 . since @xmath281 , we can assume that @xmath283 contains at least 8 vertices of @xmath57 . denote them by @xmath155 in their order in the canonical order in the boundary cycle of @xmath283 with respect to @xmath33 . by lemma [ lem : adj ] , there are @xmath291 such that @xmath288 is adjacent to @xmath158 in @xmath0 and @xmath289 is adjacent to @xmath157 . by lemmas [ lem : cons ] and [ lem : size ] , @xmath288 is not adjacent in @xmath0 to @xmath292 and @xmath293 is not adjacent to @xmath294 . notice that each of @xmath288 and @xmath289 is either laying on the boundary cycle of @xmath283 or is in another block of @xmath2 containing @xmath158 or @xmath295 and @xmath296 or @xmath157 respectively . then @xmath297 . by lemma [ lem : adj ] , there is @xmath298 such that @xmath290 is adjacent to a vertex of @xmath284 in @xmath0 . notice that @xmath299 and @xmath290 are in distinct components of @xmath258 . we obtain that @xmath290 is at distance at least 3 from @xmath288 and @xmath289 in @xmath27 . finally , assume that the vertices of @xmath57 are in the same block @xmath114 of @xmath2 . denote them by @xmath155 in their order in the canonical order in the boundary cycle of @xmath114 with respect to @xmath33 . by lemma [ lem : adj ] , there are @xmath286 such that @xmath288 is adjacent to @xmath158 in @xmath0 , @xmath289 is adjacent to @xmath300 and @xmath289 is adjacent to @xmath157 . by lemmas [ lem : cons ] and [ lem : size ] , @xmath288 is not adjacent in @xmath0 to @xmath292 , @xmath289 is not adjacent to @xmath301 and @xmath302 , and @xmath293 is not adjacent to @xmath294 . notice that each of @xmath288 , @xmath289 and @xmath289 is either laying on the boundary cycle of @xmath114 or is in another block of @xmath2 containing @xmath158 or @xmath295 , @xmath303 or @xmath300 or @xmath304 and @xmath296 or @xmath157 respectively . then @xmath305 for @xmath306 , @xmath307 . the next lemma is crucial for our algorithm . to state it , we need some additional notations . let @xmath2 be a minimal outerplanar root of a graph @xmath0 such that each vertex of @xmath2 is adjacent to at most 7 pendant vertices . let @xmath25 be a set of vertices of @xmath2 that contains all vertices of degree at least 22 . for every @xmath308 and every block @xmath114 of @xmath2 containing @xmath33 we do the following . consider the set @xmath309 and denote the vertices of @xmath57 by @xmath155 , where these vertices are numbered in the canonical order with respect to @xmath33 . then * for @xmath310 , delete the edge @xmath311 from @xmath0 if @xmath166 . * for @xmath312 , delete the edges @xmath313 from @xmath0 for @xmath314 . denote by @xmath315 the graph obtained by this procedure . [ lem : bound - tw ] there is a constant @xmath78 that depends neither on @xmath0 nor on @xmath2 such that @xmath316 let @xmath2 be a minimal outerplanar root of a graph @xmath0 such that each vertex of @xmath2 is adjacent to at most 7 pendant vertices . let @xmath25 be a set of vertices of @xmath2 containing the vertices of degree at least 22 . for each vertex @xmath308 we do the following . * let @xmath317 be the blocks of @xmath2 containing @xmath33 . * for each @xmath318 , denote by @xmath319 the neighbors of @xmath33 in @xmath287 numbered according to the canonical ordering with respect to @xmath33 for the boundary cycle of @xmath287 . assume that @xmath320 is the ordering of @xmath115 obtained by the consecutive concatenation of the sequences @xmath319 for @xmath321 . * modify @xmath2 as follows : delete @xmath33 , replace it by a path @xmath322 and make each @xmath323 adjacent to @xmath129 and @xmath130 for @xmath312 ( see figure [ fig : modif ] ) . denote by @xmath324 the graph obtained by the procedure . notice that the procedure modifies degrees of vertices of @xmath2 but we do not recompute @xmath25 . notice that @xmath2 is a contraction of @xmath324 as @xmath2 can be obtained from @xmath324 by contracting the paths @xmath322 constructed for @xmath308 . it is also straightforward to see that @xmath324 is an outerplanar graph , because each step maintains outerplanarity ( see figure [ fig : modif ] ) . by observation [ obs : tw - op ] , @xmath325 . we claim that @xmath326 . let @xmath327 . suppose first that @xmath328 . we have that @xmath329 and , in particular , @xmath34 has at most @xmath330 neighbors in @xmath25 in the graph @xmath2 . in the construction of @xmath324 each neighbor of this type is replaced by two neighbors and all other neighbors remain the same . therefore , @xmath331 . suppose now that @xmath34 is a vertex of one of the paths @xmath322 constructed for @xmath308 . notice , that when @xmath33 is replaced by @xmath322 , then the degree of each vertex @xmath323 is at most @xmath153 , and at most two neighbors of @xmath323 could be modified in the subsequent construction steps . it implies that @xmath332 . we claim that @xmath315 is a minor of @xmath333 . let @xmath334 be the graph obtained from @xmath333 by the contraction of the paths @xmath322 constructed for @xmath308 . we use the name @xmath33 for the vertex obtained by contracting @xmath322 constructed for @xmath308 . thus , we can assume that @xmath335 . we show that @xmath315 is a subgraph of @xmath334 . we already observed that @xmath2 can be obtained from @xmath324 by contracting paths @xmath322 constructed for @xmath308 . hence , each edge of @xmath2 is an edge of @xmath334 . let @xmath336 be an edge of @xmath315 that is not an edge of @xmath2 . then , there is @xmath35 such that @xmath337 . denote by @xmath338 and @xmath339 respectively the sets of vertices of @xmath324 that are contracted to @xmath88 and @xmath98 respectively . if @xmath340 , then by the construction of @xmath324 , there are @xmath341 and @xmath342 such that @xmath343 . hence , @xmath344 and @xmath345 suppose that @xmath308 . by the definition of @xmath315 , the vertices @xmath88 and @xmath98 are in the same block @xmath114 of @xmath2 . denote by @xmath346 the vertices of @xmath115 in @xmath114 in the canonical order with respect to @xmath33 along the boundary cycle of @xmath114 . we have that @xmath347 and @xmath348 for some @xmath165 . by the definition of @xmath315 , @xmath349 . by the construction of @xmath324 , there are @xmath341 and @xmath342 that are joined by the path @xmath350 in @xmath324 . since this path has length at most @xmath153 , @xmath344 . therefore , @xmath351 . since @xmath315 is a subgraph of @xmath334 and @xmath334 is a contraction of @xmath333 , we conclude that @xmath315 is a minor of @xmath333 . since , @xmath315 is a minor of @xmath333 , @xmath352 by observation [ obs : minor ] . because @xmath324 is outerplanar , @xmath325 by observation [ obs : tw - op ] , and because @xmath326 , @xmath353 by observation [ lem : tw - power ] . we obtain that @xmath354 . in this section we construct an algorithm for outerplanar root with running time @xmath17 . let @xmath0 be the input graph . clearly , it is sufficient to solve outerplanar root for connected graphs . hence , we assume that @xmath0 is connected and has @xmath355 vertices . first , we preprocess @xmath0 using lemma [ lem : pend ] to reduce the number of pendant vertices adjacent to the same vertex in a ( potential ) outerplanar root of @xmath0 . to do so , we exhaustively apply the following rule . * pendants reduction . * if @xmath0 has a set @xmath57 of simplicial true twins of size at least 8 , then delete an arbitrary @xmath356 from @xmath0 . the following claim shows that this rule is safe . [ lem : a ] if @xmath357 is obtained from @xmath0 by the application of * pendant reduction * , then @xmath0 has an outerplanar root if and only if @xmath54 has an outerplanar root . suppose that @xmath2 is a minimal outerplanar root of @xmath0 . by lemma [ lem : pend ] , @xmath2 has a pendant vertex @xmath356 . it is easy to verify that @xmath358 is an outerplanar root of @xmath54 . assume now that @xmath359 is a minimal outerplanar root of @xmath54 . by lemma [ lem : pend ] , @xmath2 has a pendant vertex @xmath360 , since the vertices of @xmath361 are simplicial true twins of @xmath54 and @xmath362 . let @xmath34 be the unique neighbor of @xmath52 in @xmath359 . we construct @xmath2 from @xmath359 by adding @xmath33 and making it adjacent to @xmath34 . it is readily seen that @xmath2 is an outerplanar root of @xmath0 . this completes the proof . for simplicity , we call the graph obtained by exhaustive application of the pendants rule @xmath0 again . the following property immediately follows from the observation that any two pendant vertices of a square root @xmath2 of @xmath0 adjacent to the same vertex in @xmath2 are true twins of @xmath0 . [ lem : b ] every outerplanar root of @xmath0 has at most @xmath168 pendant vertices adjacent to the same vertex . in the next stage of our algorithm we label some edges of @xmath0 _ red _ or _ blue _ in such a way that the edges labeled red are included in every minimal outerplanar root and the blue edges are not included in any minimal outerplanar root . we denote by @xmath363 the set of red edges and by @xmath364 the set of blue edges . we also construct a set of vertices @xmath25 of @xmath0 such that for every @xmath308 , the edges incident to @xmath33 are labeled red or blue . * labeling . * set @xmath365 , @xmath366 and @xmath367 . for each @xmath35 such that there are three distinct vertices @xmath251 that are at distance at least @xmath99 from each other in @xmath27 do the following : * set @xmath368 , * set @xmath369 , * set @xmath370 , * set @xmath371 and @xmath372 , * if @xmath373 , then return a no - answer and stop . lemmas [ lem : three ] and [ lem : high ] imply the following claim . [ lem : c ] if @xmath0 has an outerplanar root , then * labeling * does not stop in step ( v ) , and if @xmath2 is a minimal outerplanar root of @xmath0 , then @xmath374 and @xmath375 . moreover , every vertex @xmath35 with @xmath278 is included in @xmath25 . next , we find the set of edges @xmath336 with @xmath376 for some @xmath33 in @xmath363 such that they can not be edges in any minimal outerplanar root , and thus are irrelevant edges . * finding irrelevant edges . * set @xmath377 . for each @xmath308 and each pair of distinct @xmath378 such that @xmath379 do the following . * if @xmath380 , then return a no - answer and stop . * if for @xmath88 and @xmath98 , there is no @xmath381 such that @xmath382 and @xmath383 , then include @xmath336 in @xmath28 . * if @xmath384 , then return a no - answer and stop . combining lemmas [ lem : c ] and [ lem : non - edges ] , we obtain the following claim . [ lem : d ] if @xmath0 has an outerplanar root , then * finding irrelevant edges * does not stop in steps ( i ) and ( iii ) , and if @xmath2 is a minimal outerplanar root of @xmath0 , then @xmath385 . assume that we did not stop during the execution of * finding irrelevant edges. let @xmath386 . we show the following . [ lem : e ] the graph @xmath0 has an outerplanar root if and only if there is a set @xmath387 such that @xmath388 , @xmath389 , * for any @xmath390 , @xmath391 or there is @xmath392 such that @xmath393 , * for any distinct edges @xmath393 , @xmath390 or there is @xmath308 such that @xmath376 , * the graph @xmath394 is outerplanar . let @xmath2 be a minimal outerplanar root of @xmath0 . by lemma [ lem : d ] , @xmath395 , i.e. , @xmath396 . let @xmath397 . it is straightforward to verify that ( i)(iv ) are fulfilled . assume now that there is @xmath387 such that ( i)(iv ) hold . then we have that @xmath394 is an outerplanar root of @xmath0 . to complete the description of the algorithm , it remains to show how to check the existence of a set of edges @xmath398 satisfying ( i)(iv ) of lemma [ lem : e ] for given @xmath54 , @xmath363 and @xmath364 . notice , that if @xmath0 has a minimal outerplanar root @xmath2 , then @xmath54 is a subgraph of the graph @xmath315 constructed in section [ sec : tech ] by lemma [ lem : del ] . by lemma [ lem : bound - tw ] , there is a constant @xmath78 that depends neither on @xmath0 nor on @xmath2 such that @xmath399 . therefore , @xmath400 for a yes - instance . we use lemma [ l - bod ] to verify whether it holds . if we obtain that @xmath401 , we conclude that we have a no - instance and stop . otherwise , we use the celebrated theorem of courcelle @xcite , which states that any problem that can be expressed in monadic second - order logic can be solved in linear time on graph of bounded treewidth . it is straightforward to see that properties ( i)(iv ) can be expressed in this logic . in particular , to express outerplanarity in ( iv ) , we can use lemma [ lem : out - minor ] and the well - known fact that the property that @xmath0 contains @xmath114 as minor can be expressed in monadic second - order logic if @xmath114 is fixed ( see , e.g. , the book of courcelle and engelfriet @xcite ) . it immediately implies that we can decide in linear time whether @xmath398 exists or not . notice that we can modify these arguments such that we do not only check the existence of @xmath398 but also find it . to do this , we can construct a dynamic programming algorithm for graphs of bounded treewidth that finds @xmath398 . now we evaluate the running time of our algorithm . we can find simplicial vertices of the input @xmath112-vertex graph @xmath0 in time @xmath402 . therefore , * pendant reduction * can be done in time @xmath402 . for every vertex @xmath33 , we can compute the distances between the vertices of @xmath403 in @xmath27 in time @xmath402 . this implies that * labeling * can be done in time @xmath17 . * finding irrelevant edges * also can be done in time @xmath17 by checking @xmath404 pairs of vertices @xmath88 and @xmath98 . then @xmath54 can be constructed in linear time . finally , checking whether @xmath400 and deciding whether there is a set of edges @xmath398 satisfying the required properties can be done in linear time by lemma [ l - bod ] and courcelle s theorem @xcite respectively . notice that we can use the same arguments to decide whether a graph @xmath0 has a square root @xmath2 that belongs to some subclass @xmath12 of the class of outerplanar graphs . to be able to apply our structural lemmas , we only need the property that @xmath12 should be closed under edge deletions . observe also that if the properties defining @xmath12 could be expressed in monadic second - order logic , then we can apply courcelle s theorem @xcite . it gives us the following corollary . [ cor : ext ] for every subclass @xmath405 of the class of outerplanar graphs that is closed under edge deletions and can be expressed in monadic second - order logic , it can be decided in time @xmath17 whether an @xmath112-vertex graph @xmath0 has a square root @xmath406 . our main approach for solving outerplanar root is general in the sense that it can be adapted to also find square roots belonging to some other graph classes . in this section , we show that there is an algorithm to decide in polynomial time whether a graph has a square root of pathwidth at most @xmath1 . notice that graphs of pathwidth 1 are caterpillars , and that it can be decided in polynomial time whether a graph @xmath0 has a square root that is a caterpillar by an easy adaptation of the algorithms for finding square roots that are trees @xcite . we define the following problem : root of pathwidth @xmath18 : : input : : : a graph @xmath0 . question : : : is there a graph @xmath2 such that @xmath407 and @xmath84 ? the main difference between our algorithm for root of pathwidth @xmath18 and our algorithm for outerplanar root lies in the way properties of the involved graph classes are used . to show the structural results needed for this algorithm , we use the property that a potential square root has a path decomposition of width at most @xmath1 , instead of the existence of an outerplanar embedding used in the previous section . [ thm : pw2 ] root of pathwidth @xmath18 can be solved in time @xmath19 , where @xmath112 is the number of vertices of the input graph . the remaining part of this section contains the proof of theorem [ thm : pw2 ] . in section [ secpw : strc ] , we present the structural results necessary to prove correctness of the algorithm for root of pathwidth @xmath18 . the algorithm itself is presented in section [ secpw : algo ] . in this section , we present structural properties of graphs with square roots of pathwidth at most @xmath1 . recall that we assume that for any considered path decomposition , it holds that @xmath408 for every two distinct bags @xmath279 and @xmath364 of the decomposition . as the pathwidth of any subgraph of a graph of pathwidth at most @xmath1 is upper bounded by 2 , by observation [ obs : min ] , we can consider only minimal roots of pathwidth at most @xmath1 . [ lem : neighxi ] let @xmath0 be a graph and @xmath2 be a minimal square root of @xmath0 such that @xmath407 . suppose there exist distinct @xmath409 such that the bags @xmath410 , @xmath411,@xmath412 , @xmath413 appear in this order in the path decomposition of @xmath2 . then for each @xmath414 , @xmath415 , @xmath416 . suppose @xmath130 has a neighbor @xmath52 in @xmath2 such that @xmath417 and @xmath418 . clearly there exists a bag @xmath364 in the path decomposition of @xmath2 that contains @xmath130 and @xmath52 . as @xmath364 contains @xmath130 , then @xmath364 is between the bags @xmath419 and @xmath420 in the path decomposition and hence must contain @xmath33 and @xmath34 . then @xmath421 , a contradiction with @xmath407 . for two positive integers @xmath422 and @xmath423 , we denote by @xmath424 the _ ramsey number _ , that is , the smallest @xmath112 such that every graph on @xmath112 vertices has either a clique of size @xmath422 or an independent set of size @xmath423 . those numbers are all finite by ramsey s theorem ( see , e.g. , @xcite ) . [ lem : pwreduction1 ] let @xmath0 be a graph and @xmath2 be a minimal square root of @xmath0 such that @xmath407 . then there is a constant @xmath425 such that for every set @xmath426 of true twins in @xmath0 , if @xmath427 , then one of the following holds : * @xmath426 has a vertex that is a pendant in @xmath2 . * @xmath426 has distinct pairwise nonadjacent vertices @xmath428 of degree 2 in @xmath2 such that @xmath429 . let @xmath430 . we start by constructing an auxiliary graph @xmath114 . let @xmath431 and two vertices of @xmath114 are adjacent if there exists a bag in the path decomposition of @xmath2 that contains both of them . we want to show that @xmath114 has an independent set of size 16 . to do so , we prove that @xmath114 has no clique of @xmath153 vertices . suppose for the sake of contradiction that @xmath114 has such a clique , say @xmath432 . let @xmath433 be the interval of those bags of the path decomposition of @xmath2 that contain the vertex @xmath130 . by the construction of @xmath114 , these intervals are pairwise intersecting . by the helly property , there exists a bag containing all four vertices , a contradiction with @xmath407 . hence @xmath114 has an independent set of size 16 , since @xmath434 and @xmath114 has no clique of @xmath153 vertices . let @xmath435 be such an independent set of @xmath114 . by the construction of @xmath114 , there are no two vertices of @xmath436 that are contained in the same bag of the path decomposition of @xmath2 . let @xmath437 be bags containing the vertices of @xmath436 such that @xmath438 . as the vertices of @xmath436 are true twins in @xmath0 and they are not adjacent in @xmath2 , there must exist a path of length two between any two of them . let @xmath439 be such path between @xmath158 and @xmath440 . we may consider that @xmath441 , otherwise we can pick bag @xmath442 as the one containing @xmath158 and @xmath33 . analogously , we may assume that @xmath443 . since we have a path decomposition , we conclude that @xmath444 , @xmath445 . suppose there are three distinct vertices @xmath130 , @xmath228 and @xmath157 in @xmath436 that are not adjacent to @xmath33 , with @xmath446 . let @xmath34 be a vertex in a path of length two between @xmath130 and @xmath157 , i.e. @xmath34 is a common neighbor of @xmath130 and @xmath157 . we may assume that @xmath447 and @xmath448 . then @xmath449 . by lemma [ lem : neighxi ] and by the fact that @xmath450 , we conclude that @xmath451 and hence @xmath452 holds . suppose that at most two vertices of @xmath453 are not adjacent to @xmath33 in @xmath2 . let @xmath454 , @xmath455 , be the set containing the vertices that are adjacent to @xmath33 in @xmath453 . if some vertex of @xmath456 has degree one in @xmath2 , then condition @xmath452 holds . assume this is not the case . then all the vertices of @xmath456 have degree at least two in @xmath2 . let @xmath457 be bags such that @xmath458 . let @xmath34 be a neighbor of @xmath459 , @xmath460 . we may assume @xmath461 . if @xmath34 appears in at least five bags of @xmath462 , then it appears in @xmath463 , @xmath464 , @xmath465 , @xmath466 and @xmath467 . as @xmath468 , @xmath469 and @xmath470 do not have degree one by hypothesis , then by lemma [ lem : neighxi ] , they have degree two and hence condition @xmath471 holds . as @xmath459 and @xmath440 are true twins in @xmath0 , then @xmath34 must also be a neighbor of @xmath440 in @xmath0 . if @xmath34 and @xmath440 are not adjacent , let @xmath472 be a path of length two between them . if @xmath417 , the bag containing @xmath473 has to be between @xmath474 and @xmath475 . since @xmath455 , either @xmath34 or @xmath52 appear in at least five bags of @xmath462 . if @xmath476 , then @xmath34 , @xmath33 and @xmath459 form a triangle in @xmath2 . by observation [ obs : triangle ] , there exists a vertex @xmath126 that is adjacent to either @xmath34 or @xmath459 that is not adjacent to @xmath33 . the case when @xmath126 is a neighbor of @xmath459 has already been covered . suppose @xmath477 and @xmath478 . as @xmath479 , z is adjacent to @xmath459 in @xmath0 , meaning that @xmath126 must also be adjacent to @xmath440 in @xmath0 . if @xmath480 , then the bag @xmath481 appears between bags @xmath463 and the bag containing @xmath440 and @xmath126 . then either @xmath34 or @xmath126 appear in at least five bags of @xmath462 . if @xmath482 , let @xmath483 be a path of length two in @xmath2 . since @xmath455 and there must exist a bag containing @xmath440 and @xmath98 , and thus one of @xmath34 , @xmath126 or @xmath98 appear in at least five bags of @xmath462 . if @xmath0 has a set @xmath426 of true twins with size greater than @xmath425 , then we can delete an arbitrary vertex of @xmath426 . the following lemma shows that this procedure is safe . [ lem : pwreduction2 ] let @xmath426 be a set of true twins in @xmath0 such that @xmath484 and let @xmath485 . then @xmath0 has a square root @xmath2 such that @xmath407 if and only if @xmath357 has a square root @xmath359 such that @xmath486 . let @xmath0 be a graph that has a square root @xmath2 of pathwidth at most @xmath1 . since @xmath484 , either there exists a vertex @xmath487 satisfying the condition ( i ) of lemma [ lem : pwreduction1 ] or there are three vertices @xmath488 satisfying ( ii ) . since the vertices of @xmath426 are true twins , we can assume that @xmath489 in the first case and @xmath490 in the second . in both cases we obtain that @xmath258 is a square root of pathwidth at most @xmath1 for @xmath27 . to prove the other direction , let @xmath359 be a square root of pathwidth at most @xmath1 of @xmath54 . since @xmath491 , @xmath359 has a vertex @xmath34 satisfying the condition ( i ) of lemma [ lem : pwreduction1 ] or there are three vertices @xmath488 satisfying ( ii ) . if @xmath34 is a pendant , let @xmath52 be the vertex of @xmath359 that is adjacent to @xmath34 . in order to obtain a square root @xmath2 of @xmath0 , we attach @xmath33 to @xmath52 in @xmath359 . we now prove that @xmath407 . we may assume that @xmath34 appears in only one bag in the path decomposition of @xmath359 ( which also contains @xmath52 ) . if this is not the case , we can delete all other occurrences of @xmath34 and still obtain a valid path decomposition of @xmath359 having width at most two . let @xmath492 be the bag containing @xmath493 and @xmath494 be the next bag in the path decomposition . if @xmath495 , we create a new bag between @xmath492 and @xmath494 containing @xmath496 . if @xmath497 , then @xmath498 , and the new bag will contain @xmath499 . this is a path decomposition for @xmath2 of width at most two . assume that @xmath261 are vertices satisfying condition ( ii ) of lemma [ lem : pwreduction1 ] . let @xmath500 . in order to obtain a square root @xmath2 for @xmath0 , we add @xmath33 to @xmath501 and make it adjacent to @xmath52 and @xmath98 . we now prove that @xmath407 . since @xmath500 , there is a bag in the path decomposition of @xmath359 containing @xmath502 for some @xmath253 . since @xmath7 is only adjacent to @xmath52 and @xmath98 , we may assume that this is the only bag in the decomposition containing @xmath7 . let @xmath494 be the bag that succeeds @xmath492 in the decomposition . we create a new bag between @xmath492 and @xmath494 containing @xmath503 , and hence obtain a path decomposition for @xmath2 of width at most two . for simplicity , let us denote the graph obtained by exhaustive application of the previously described safe procedure by @xmath0 again . from now on , in all our lemmas , we assume that every set of true twins of @xmath0 has size bounded by the constant @xmath425 . [ lem : pwboundxi ] let @xmath0 be a graph and @xmath2 be a minimal square root of @xmath0 such that @xmath407 . suppose there exist distinct @xmath409 such that the bags @xmath410 , @xmath411,@xmath412 , @xmath413 appear in this order in the path decomposition of @xmath2 . then @xmath504 . by lemma [ lem : neighxi ] , for every @xmath415 , @xmath416 . all vertices that are adjacent in @xmath2 only to @xmath33 are true twins in @xmath0 . the same applies to vertices that are adjacent only to @xmath34 and also to those that are adjacent to both @xmath33 and @xmath34 . since the size of every set of true twins in @xmath0 is bounded by @xmath425 , we have @xmath505 and hence @xmath504 . [ lem : pwcommonneigh ] let @xmath0 be a graph and @xmath2 be a minimal square root of @xmath0 such that @xmath407 . if @xmath506 , then the number of common neighbors of @xmath33 and @xmath34 in @xmath2 is bounded by @xmath507 . suppose @xmath508 . in the path decomposition of @xmath2 , for each @xmath414 , we need a bag containing @xmath33 and @xmath130 and a bag containing @xmath34 and @xmath130 . as @xmath407 , this implies the existence of one bag @xmath509 for each @xmath414 . suppose @xmath510 appear in the path decomposition in this order . lemma [ lem : neighxi ] and the fact that @xmath511 imply that the vertices @xmath512 are true twins in @xmath0 and that the number of such vertices is bounded by @xmath425 . hence , the number of common neighbors of two vertices in @xmath2 is bounded by @xmath507 . [ lem : pwfive ] let @xmath0 be a graph and @xmath2 be a minimal square root of @xmath0 such that @xmath407 . let @xmath35 and @xmath513 . if @xmath514 , then there exist five vertices @xmath515 in @xmath403 such that @xmath516 for all @xmath517 . let @xmath35 be such that @xmath514 . we choose a set of bags @xmath518 in the path decomposition of @xmath2 such that @xmath444 . then for all @xmath414 and for all @xmath144 , there exists an @xmath414 such that @xmath519 . note that some neighbors of @xmath33 might appear in more than one bag . let @xmath520 be the smallest integer such that @xmath521 contains at least three distinct vertices of @xmath115 . since @xmath33 belongs to all bags , at least one of these three neighbors in @xmath521 does not appear in @xmath522 . let @xmath288 be such a vertex . in general , let @xmath523 be the smallest integer greater than @xmath524 such that @xmath525 contains at least five new vertices of @xmath115 . since @xmath33 belongs to all bags , there is at least one vertex among the five that appears neither in @xmath526 nor in @xmath527 . let @xmath8 be such a vertex . hence we obtain a set @xmath528 that is an independent set . since @xmath529 , we have @xmath530 . for each @xmath7 , we pick @xmath531 such that @xmath532 if @xmath307 . since @xmath530 and by lemma [ lem : pwcommonneigh ] , we can pick at least 21 such vertices . for simplicity , we assume that @xmath533 is a set of such vertices . for each @xmath414 , let @xmath492 be a bag containing @xmath7 and @xmath33 . for @xmath534 , we can assume @xmath535 . note that @xmath130 and @xmath536 might be adjacent in @xmath2 , but @xmath130 can not be a neighbor of @xmath157 , with @xmath537 , because of the existence of bag @xmath538 . for the same reason , @xmath130 can not be adjacent to @xmath293 for some @xmath537 . also , if @xmath537 , all paths in @xmath2 from @xmath130 to @xmath157 contain either @xmath536 or @xmath539 . the same applies to the paths from @xmath130 to @xmath293 for some @xmath537 . then @xmath540 are vertices that are pairwise at distance at least 3 in @xmath27 . [ lem : pwlabelu ] let @xmath0 be a graph and @xmath2 be a minimal square root of @xmath0 such that @xmath407 . if @xmath0 has five vertices @xmath541 such that @xmath542 , then for any @xmath252 , @xmath138 if and only if there exists an @xmath543 such that @xmath544 . by lemma [ lem : not - incl ] , if there exists @xmath543 such that @xmath544 , then @xmath138 . let @xmath252 be such that @xmath138 . we need to prove that there exists an @xmath414 such that @xmath544 . this holds if @xmath545 for some @xmath546 . note also that @xmath547 because @xmath542 . then for each @xmath414 there exists @xmath7 such that @xmath548 and @xmath549 . if @xmath307 , then @xmath550 , otherwise @xmath130 and @xmath228 would be adjacent in @xmath0 moreover , @xmath551 , otherwise @xmath552 . without loss of generality assume that @xmath553 appear in this order in the path decomposition of @xmath2 . this implies the existence of the bags @xmath554 , @xmath555 and @xmath556 in the path decomposition of @xmath2 . consider the bags containing @xmath88 . if @xmath88 appears before @xmath289 , consider the distance between @xmath88 and @xmath248 . if the shortest path between @xmath88 and @xmath248 in @xmath27 contains @xmath295 or @xmath156 , then @xmath557 . if it does not contain @xmath295 and @xmath156 , then it must contain either @xmath289 , @xmath290 or another neighbor of @xmath33 that appeared previously in the path decomposition and has no common neighbor with @xmath248 in @xmath2 . however , @xmath558 , otherwise @xmath559 . as @xmath558 , we have @xmath560 . if @xmath88 appears between @xmath554 and @xmath555 , consider @xmath561 . by the same argument we obtain that @xmath562 , because of the existence of bags @xmath555 and @xmath556 . the other cases follow by symmetry . given a vertex @xmath35 such that @xmath514 , we define @xmath563 and @xmath564 . let @xmath565 . if @xmath566 , by observation [ obs : triangle ] , there exists a vertex @xmath34 that is adjacent to at least one of @xmath567 and @xmath568 in @xmath2 , but is not adjacent to @xmath33 . assume @xmath34 is adjacent to @xmath567 . this implies @xmath569 . moreover , as @xmath566 , we have @xmath570 . in other words , if there exists an edge in @xmath2 between two vertices in @xmath571 , then there exists a vertex in @xmath572 that is adjacent to both of them in @xmath0 . we can now use this observation to identify edges that will not be in a square root of @xmath0 . let @xmath569 and @xmath573 . we formalize this in the following lemma . [ lem : pwexclude ] let @xmath35 be such that @xmath514 and @xmath113 for a minimal square root @xmath2 of @xmath0 with @xmath407 . if there is no @xmath569 such that @xmath574 , then @xmath140 . [ lem : c3 ] let @xmath0 be a graph and @xmath2 a minimal square root of @xmath0 such that @xmath407 . let @xmath33 be a vertex such that @xmath514 and @xmath569 . let @xmath279 and @xmath575 be , respectively , the first and the last bag in the path decomposition of @xmath2 containing @xmath33 and a vertex of @xmath576 . there exists a constant @xmath577 such that the number of bags between @xmath279 and @xmath575 is bounded by @xmath578 . let @xmath279 and @xmath575 be the first and the last bag in the path decomposition of @xmath2 containing @xmath33 and a vertex of @xmath576 . if @xmath34 belongs to both @xmath279 and @xmath575 , then all the bags between @xmath279 and @xmath575 ( and including them ) contain both @xmath33 and @xmath34 . since for any two bags @xmath579 in the path decomposition we have @xmath580 , for every vertex appearing between @xmath279 and @xmath575 we have exactly one new bag . the number of such vertices is bounded by @xmath581 , by lemma [ lem : pwboundxi ] . if @xmath34 appears before @xmath279 and is not contained in @xmath279 , let @xmath582 be such that @xmath583 . since the bags containing @xmath34 appear before @xmath279 , @xmath88 can not be adjacent to @xmath34 in @xmath2 . by the definition of @xmath576 , they are at distance two . let @xmath584 be such that @xmath585 . there exists a bag containing @xmath586 , which is before @xmath279 in the path decomposition and a bag containing @xmath587 , which is between @xmath279 and @xmath575 since @xmath582 . the number of bags between @xmath279 and @xmath587 is bounded by @xmath581 , by lemma [ lem : pwboundxi ] , as both contain @xmath98 and @xmath33 . the same holds for the number of bags between @xmath587 and @xmath575 , as they both contain @xmath88 and @xmath33 . if @xmath34 appears in @xmath279 but not in @xmath588 , let @xmath582 be such that @xmath583 . since @xmath589 , let @xmath590 be a path between @xmath88 and @xmath34 in @xmath2 . there exists a bag containing @xmath591 and a bag containing @xmath592 that appears after @xmath591 . the same constant @xmath581 bounds the number of bags between @xmath279 and @xmath591 , between @xmath591 and @xmath592 and between @xmath592 and @xmath575 . if @xmath34 appears between @xmath279 and @xmath575 but it is not contained in them , we proceed in the same way as we did in the previous cases and the worst scenario is the one where the vertices of @xmath576 contained in @xmath279 and @xmath575 are not adjacent to @xmath34 . let @xmath593 and @xmath584 be such that @xmath590 is a path between @xmath88 and @xmath34 . we take @xmath594 and @xmath595 analogously with respect to @xmath575 . by lemma [ lem : pwboundxi ] , the constant @xmath581 bounds the number of bags between the following pairs of bags : @xmath279 and @xmath587 , @xmath587 and @xmath591 , @xmath591 and @xmath596 , @xmath596 and @xmath597 and finally , @xmath597 and @xmath575 . the total number of bags between @xmath279 and @xmath575 is therefore bounded by @xmath578 . let @xmath2 be a minimal square root of @xmath0 with @xmath407 . let @xmath598 be the subgraph of @xmath0 obtained by deleting the edges between @xmath113 , where @xmath514 , such that there is no @xmath569 such that @xmath574 . [ lem : pwbdtw ] there exists a constant @xmath599 that depends neither on @xmath0 nor on @xmath2 such that @xmath600 . let @xmath2 be a minimal square root of @xmath0 . let @xmath25 be the set of vertices of @xmath2 such that @xmath514 for each @xmath308 , we do the following . consider the bags @xmath601 in the path decomposition of @xmath2 containing @xmath33 and its neighbors . starting from @xmath442 , we pick the first bag where a new neighbor of @xmath33 appears . let @xmath602 be such a bag . as @xmath602 contains at least one vertex that is not contained in @xmath603 , we have @xmath604 and we already know that @xmath605 . in the bags @xmath606 , we replace @xmath33 by @xmath567 . we create a new bag between @xmath603 and @xmath602 containing @xmath567 , @xmath568 and @xmath607 . in the bags @xmath608 , we replace @xmath33 by @xmath568 . in general , for every bag @xmath609 found containing a new neighbor of @xmath33 we do the following : 1 . create a new bag between @xmath610 and @xmath609 containing @xmath611 and the vertices of @xmath612 ( note that @xmath613 ) . 2 . in the bags @xmath614 , replace @xmath615 by @xmath611 . 3 . in @xmath2 , add the edge between @xmath615 and @xmath611 and an edge between @xmath611 and the new found neighbor of @xmath33 . let @xmath324 be the graph obtained from @xmath2 by the previous procedure . note that @xmath2 is a contraction of @xmath324 , as @xmath2 can be obtained by contracting the edges of the paths created for each vertex of @xmath25 . since we construct a path decomposition for @xmath324 with the same width as the one we had for @xmath2 , we have @xmath616 . we claim that @xmath617 is bounded . if @xmath328 , then @xmath34 has bounded degree in @xmath2 and , in each step of the above procedure , the degree of @xmath34 is maintained . the vertices @xmath323 created for each vertex of @xmath25 have degree bounded by three . thus the graph @xmath324 has indeed bounded degree . we claim that @xmath598 is a minor of @xmath618 . let @xmath334 be obtained from @xmath618 by the contraction of the paths created for each vertex of @xmath25 . we can now assume that @xmath619 . we now show that @xmath598 is a subgraph of @xmath334 . every edge of @xmath598 that belongs to @xmath620 is also an edge of @xmath618 . let @xmath621 be such that @xmath140 . as @xmath2 is a square root of @xmath0 , there exists @xmath622 such that @xmath169 . denote by @xmath338 and @xmath339 respectively the sets of vertices of @xmath324 that were contracted to @xmath88 and @xmath98 . if @xmath340 , by the construction of @xmath324 , there are @xmath341 and @xmath342 such that @xmath343 and therefore @xmath623 and @xmath345 . if @xmath308 , there exists a path @xmath624 in @xmath324 , @xmath625 and @xmath626 such that @xmath627 and @xmath628 . since @xmath621 and @xmath308 , we know that @xmath574 for some @xmath34 , otherwise we would have deleted this edge in the construction of @xmath598 . as the number of bags containing @xmath33 and vertices of @xmath576 is bounded by @xmath577 by lemma [ lem : c3 ] , the length of the path @xmath624 is at most @xmath577 . this implies that @xmath629 and hence @xmath630 , which implies @xmath631 . since @xmath598 is a subgraph of @xmath334 and @xmath334 is a contraction of @xmath618 , we conclude that @xmath598 is a minor of @xmath618 . since @xmath598 is a minor of @xmath618 , @xmath632 . as @xmath324 has pathwidth at most @xmath1 , @xmath616 . since @xmath324 has bounded degree , by observation [ lem : tw - power ] , we obtain a constant @xmath599 such that @xmath600 . let @xmath0 be the input graph . it is sufficient to solve root of pathwidth @xmath633 for connected graphs . hence , we assume that @xmath0 is connected and has @xmath355 vertices . first , we preprocess @xmath0 using lemma [ lem : pwreduction1 ] to reduce the number of true twins that a given vertex of @xmath20 might have . to do so , we exhaustively apply the following rule , which is safe by lemma [ lem : pwreduction2 ] . * twin reduction . * if @xmath0 has a set @xmath57 of true twins of size at least @xmath634 , then delete an arbitrary @xmath356 from @xmath0 . for simplicity , we call the graph obtained by exhaustive application of the twin reduction rule @xmath0 again . in the next stage of our algorithm we label some edges of @xmath0 _ red _ or _ blue _ in such a way that the edges labeled red are included in every minimal square root of pathwidth at most @xmath1 and the blue edges are not included in any minimal square root of pathwidth at most @xmath1 . we denote by @xmath363 the set of red edges and by @xmath364 the set of blue edges . we also construct a set of vertices @xmath25 of @xmath0 such that for every @xmath308 , the edges incident to @xmath33 are labeled red or blue . * labeling . * set @xmath365 , @xmath366 and @xmath367 . for each @xmath35 such that there are five distinct vertices in @xmath403 that are at distance at least @xmath99 from each other in @xmath27 do the following : * set @xmath368 , * set @xmath635 , * set @xmath370 , * set @xmath371 and @xmath372 , * if @xmath373 , then return a no - answer and stop . lemmas [ lem : pwfive ] and [ lem : pwlabelu ] imply the following claim . if @xmath0 has a square root of pathwidth at most @xmath1 , then * labeling * does not stop in step ( v ) . moreover , @xmath374 , @xmath375 and every vertex @xmath35 with @xmath514 is included in @xmath25 . next , we find the set of edges @xmath336 with @xmath376 for some @xmath33 in @xmath25 that are not an edge in any square root of pathwidth at most @xmath1 . * finding irrelevant edges . * set @xmath377 . for each @xmath308 and each pair of distinct @xmath378 such that @xmath379 and @xmath103 do the following . * if there is no @xmath381 such that @xmath382 and @xmath636 , then include @xmath336 in @xmath28 . * if @xmath384 , then return a no - answer and stop . by lemma [ lem : pwexclude ] , we obtain the following claim . [ lem : pwexclude2 ] if @xmath0 has a square root of pathwidth at most @xmath1 , then * finding irrelevant edges * does not stop in step ( ii ) , and if @xmath2 is such a square root of @xmath0 , then @xmath385 . assume that we did not stop during the execution of * finding irrelevant edges. let @xmath386 . we show the following . [ lem : pwsq ] the graph @xmath0 has a square root of pathwidth at most @xmath1 if and only if there is a set @xmath387 such that @xmath388 , @xmath389 , * for any @xmath390 , @xmath391 or there is @xmath392 such that @xmath393 , * for any distinct edges @xmath393 , @xmath390 or @xmath637 , * the graph @xmath394 is such that @xmath407 . let @xmath2 be a square root of pathwidth at most @xmath1 of @xmath0 . by lemma [ lem : pwexclude2 ] , @xmath395 , i.e. , @xmath396 . let @xmath397 . it is straightforward to verify that ( i)(iv ) are fulfilled . assume now that there is @xmath387 such that ( i)(iv ) hold . then we have that @xmath394 is a square root of pathwidth at most @xmath1 of @xmath0 . to complete the description of the algorithm , it remains to show how to check the existence of a set of edges @xmath398 satisfying ( i)(iv ) of lemma [ lem : pwsq ] for given @xmath54 , @xmath363 , @xmath364 and @xmath28 . if @xmath0 has a minimal square root @xmath2 with @xmath407 , then @xmath54 is a subgraph of @xmath598 by lemmas [ lem : pwfive ] and [ lem : pwexclude ] . then by lemma [ lem : pwbdtw ] , @xmath638 therefore , @xmath638 for a yes - instance . by lemma [ l - bod ] , it is possible to verify in linear time whether it holds . if @xmath639 , we conclude that we have a no - instance and stop . otherwise , notice that properties ( i ) ( iv ) can be expressed in monadic second order logic . in particular , ( iv ) can be expressed using the property that the class of graphs of pathwidth at most @xmath1 is defined by the set of forbidden minors given by kinnersley and langston in @xcite . then we use courcelle s theorem @xcite to decide in linear time whether @xmath398 exists or not . now we evaluate the running time of our algorithm . we can find classes of true twin vertices of the input @xmath112-vertex graph @xmath0 in time @xmath402 . therefore , * twin reduction * can be done in time @xmath402 . for every vertex @xmath33 , we can compute the distances between the vertices of @xmath403 in @xmath27 in time @xmath402 . it implies that * labeling * can be done in time @xmath19 . * finding irrelevant edges * can be done in time @xmath17 by checking @xmath404 pairs of vertices @xmath88 and @xmath98 . then @xmath54 can be constructed in linear time . finally , checking whether @xmath638 and deciding whether there is a set of edges @xmath398 satisfying the required properties can be done in linear time by lemma [ l - bod ] and courcelle s theorem @xcite respectively . this implies that the total running time is also @xmath19 . we proved that @xmath11-square root is polynomial - time solvable when @xmath11 is the class of outerplanar graphs or the class of graphs of pathwidth at most 2 . the same result holds if @xmath640 is any subclass of the class of outerplanar graphs that is closed under edge deletion and that can be expressed in monadic second - order logic ( for instance , if @xmath12 is the class of cactus graphs ) . we conclude by posing two questions : * is @xmath11-square root polynomial - time solvable for every class @xmath12 of graphs of bounded pathwidth ? * is @xmath11-square root polynomial - time solvable if @xmath11 is the class of planar graphs ? cochefert , m. , couturier , j. , golovach , p.a . , kratsch , d. , paulusma , d. : sparse square roots . in : graph - theoretic concepts in computer science - 39th international workshop , wg 2013 . lncs , vol . 8165 , pp . springer ( 2013 ) courcelle , b. , engelfriet , j. : graph structure and monadic second - order logic - a language - theoretic approach , encyclopedia of mathematics and its applications , vol . cambridge university press ( 2012 ) golovach , p.a . , kratsch , d. , paulusma , d. , stewart , a. : finding cactus roots in polynomial time . in : proceedings of the 27th international workshop on combinatorial algorithms ( iwoca 2016 ) . lncs , vol . 9843 , pp . 253265 ( 2016 ) golovach , p.a . , kratsch , d. , paulusma , d. , stewart , a. : a linear kernel for finding square roots of almost planar graphs . in : proceedings of the 15th scandinavian symposium and workshops on algorithm theory ( swat 2016 ) . leibniz international proceedings in informatics , vol . 53 , pp . 4:14:14 ( 2016 ) golovach , p.a . , kratsch , d. , paulusma , d. , stewart , a. : squares of low clique number . in : 14th cologne twente workshop 2016 ( ctw 2016 ) . electronic notes in discrete mathematics , vol . 55 , pp . 195198	deciding whether a given graph has a square root is a classical problem that has been studied extensively both from graph theoretic and from algorithmic perspectives . the problem is np - complete in general , and consequently substantial effort has been dedicated to deciding whether a given graph has a square root that belongs to a particular graph class . there are both polynomial - time solvable and np - complete cases , depending on the graph class . we contribute with new results in this direction . given an arbitrary input graph @xmath0 , we give polynomial - time algorithms to decide whether @xmath0 has an outerplanar square root , and whether @xmath0 has a square root that is of pathwidth at most @xmath1 .
in 1974 hawking showed that black holes should continually emit radiation as a natural consequence of quantum field theory in curved spacetime @xcite . hawking radiation is analogous to thermal radiation from a black body and leads to the identification of a classical black hole temperature . in the case of a schwarzschild black hole , this temperature is inversely proportional to the black hole mass and therefore increases as the black hole radiates @xcite . only black holes with mass below @xmath2 have a temperature today greater than that of the cosmic microwave background . such black holes may have been produced in the early universe with dramatic cosmological consequences . formation scenarios for primordial black holes ( pbhs ) include the collapse of overdense regions due to primordial inhomogeneities , especially those generated by inflation , a softening of the equation of state or bubble collisions at cosmological phase transitions , and the collapse of oscillating cosmic string loops or domain walls . the cosmological consequences include modifying the usual picture of cosmological nucleosynthesis , distorting the spectrum of the microwave background radiation , generating a cosmic baryon asymmetry , removing magnetic monopoles , and contributing to the dark matter . even if pbhs never existed , studying them therefore places important constraints on models of the early universe . all of these formation scenarios and cosmological consequences have recently been reviewed by carr @xcite . of particular interest are those pbhs with initial mass @xmath3 which are evaporating at the present epoch , since these should contribute to the observable @xmath4-ray , cosmic ray and neutrino backgrounds ( @xcite and references therein ) . indeed the detection of such backgrounds or of individual pbh bursts would be the first confirmation of the fundamental relationship linking black holes , general relativity and thermodynamics ( see also ref . @xcite ) . to date , no such pbh burst or background has been conclusively detected , although d. cline et al . @xcite have attributed some short - timescale gamma - ray bursts to pbh explosions . however , observations of the 100 mev gamma - ray background imply that @xmath5 pbhs must have @xmath6 @xcite , where @xmath7 is their present density in units of the critical density . the 5-year data from the whipple gamma - ray telescope impose the strongest current limit on the pbh explosion rate @xcite . with the standard model of particle physics , this is much weaker than the gamma - ray background limit on @xmath7 if the pbhs cluster smoothly within the galactic halo , but it could be stronger if pbhs were highly clumped . the @xmath7 limits from the galactic cosmic rays are comparable with the gamma - ray background limit but depend on both the pbh clustering and the cosmic ray propagation model @xcite . black holes of @xmath3 have an initial temperature of about 20 mev . because a black hole will emit all particle species with rest mass less than or of order the black hole temperature , the emission from a 20 mev black hole is sensitive to the threshold for quark and gluon production , @xmath8 . in the standard hawking model @xcite , the particles directly evaporated from a black hole are those which appear noncomposite compared to the wavelength of the radiated energy ( or equivalently the black hole size ) at a given temperature . these particles can then decay or form composite particles as the emission streams away from the black hole . a black hole with a temperature somewhat below @xmath1 should directly emit neutral and charged pions , as well as photons , electrons , positrons , muons , neutrinos and gravitons . once the energy distribution of the hawking emission significantly extends above @xmath9 , the black hole should directly emit quarks and gluons rather than pions , with the quarks and gluons fragmenting and hadronizing in jets into the astrophysically stable species ( @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 and @xmath16 ) after emission . this process is analogous to qcd jet decay in accelerators . by convolving the hawking evaporation with the herwig monte carlo jet code , macgibbon and webber found that the particle flux from @xmath17 black holes is predominantly comprised of jet fragmentation products @xcite . they found that , after decay , the flux peaks at an energy of around 100 mev , roughly independent of the black hole temperature , with the average energy of the final species scaling more weakly than the black hole temperature . in contrast , prior to decay , the peak and average energies of the direct hawking emission are proportional to the black hole temperature . in the heckler photosphere scenario , as originally described in refs . @xcite , the situation is radically different . above a certain black hole temperature , the particles are alleged to interact sufficiently after emission to form a quasithermal photosphere at a distance @xmath18 times the schwarzschild radius from the black hole @xcite , where @xmath19 is the fine structure constant . in this model , the relevant qed interactions are bremsstrahlung and electron - photon pair production . the critical temperature above which this happens is estimated to be @xmath20 gev . when qcd effects are included , a lower transition temperature @xmath21 is derived where @xmath22 is the strong coupling constant : @xmath23 at energy scales around 100 gev but it increases at lower energies and its precise behavior is unknown around @xmath1 . once the photosphere develops , the high energy emission is reprocessed to lower energies , drastically reducing the flux of high energy particles and significantly weakening the detectability of individual high - temperature black holes . because the best limits on @xmath7 are derived from the present 100 mev emission from a galactic or extragalactic background of pbhs with temperatures well below @xmath24 , the limits on @xmath7 are weakened only slightly by photosphere development @xcite . however , photosphere development would have considerable implications for experimental efforts to detect ultra - high - energy cosmic rays from the final explosive phases of pbhs in the present epoch @xcite . indeed , this has led to decreased motivation among groups searching for such signatures . j. cline , mostoslavsky , and servant @xcite have numerically explored the development of the photosphere in the original heckler model . they conclude that photosphere formation should set in above a slightly lower value of @xmath24 , weakening the bound on @xmath7 still further and significantly changing the burst spectra . these authors , however , assume the same initial description of interactions as heckler @xcite , so their work is not an independent derivation or confirmation of the model . kapusta @xcite and daghigh and kapusta @xcite have also performed calculations related to heckler s model . in this paper we will raise a number of problems with the heckler scenario . these include various geometrical effects , the form of the plasma mass correction , the validity of the perfect fluid assumption and qcd phenomenological considerations . however , the two most serious omissions in the heckler scenario are as follows : ( i ) two particles must be in causal contact to interact , and this is satisfied only by a small fraction of the emitted relativistic particles when they are sufficiently close to the black hole that they would otherwise have significant interactions ; and ( ii ) in a bremsstrahlung interaction the electron initially forms off - shell and must travel a distance in the black hole frame much greater than @xmath25 before completing each interaction and becoming on - shell . if it is ` bumped ' by encounters with other particles within @xmath25 of the hole , the landau - pomeranchuk - migdal ( lpm ) effect implies that the particle generally undergoes at most one completed bremsstrahlung interaction as it streams away from the black hole . thus we conclude that a persistent qed or qcd photosphere can not form around a black hole . rare interactions could slightly modify the @xmath10 , @xmath11 , @xmath12 , @xmath26 , @xmath27 , @xmath15 and @xmath16 signatures of pbhs but not significantly . other authors , such belyanin , kocharovsky , and kocharovsky @xcite and d. cline et al . @xcite , have proposed photosphere models which are unrelated to the heckler mechanism . these models are mainly motivated by an attempt to explain certain short - period gamma - ray bursts . after analyzing the belyanin - kocharovsky - kocharovsky and d. cline et al . models , we conclude that particle interactions are also insufficient to form a qed or qcd photosphere in these scenarios . however , the relevance of pbh explosions at the present epoch should not be ruled out as some other pbh mechanism which has not yet been considered might explain the observed bursts . the plan of this paper is as follows . in section ii we review the fundamentals of hawking evaporation . in section iii we outline the original heckler photosphere scenario . in section iv we analyze this scenario in detail . in section v we summarize the results of our analysis and their observational implications . other photosphere models are discussed in section vi . our conclusions are presented in section vii . except in sections ii and via , we will use planck units in which @xmath28 . a kerr - newman black hole with angular velocity @xmath29 and electric potential @xmath30 radiates particles with total energy between @xmath31 at a rate @xcite @xmath32-\left(-1\right)^{2s } \right]^{-1 } \label{1}\ ] ] per degree of particle freedom . here @xmath33 is the particle spin , @xmath34 is its axial quantum number or angular momentum , @xmath35 is its orbital angular momentum , @xmath36 is its charge and @xmath37 is the dimensionless absorption probability for the emitted species . in general , @xmath37 is a function of @xmath38 , @xmath29 , @xmath30 and the surface gravity of the black hole @xmath39 , as well as the internal degrees of freedom and rest mass of the emitted particle . equation implies that a charged , rotating black hole will preferentially emit particles which have the same sign of charge and spin as the black hole . the angular momentum of a black hole is emitted several times faster than its mass @xcite . an electrically charged black hole quickly discharges provided @xmath40 , where @xmath41 is the black hole mass @xcite . thus we will henceforth assume an uncharged , nonrotating schwarzschild black hole , i.e. @xmath42 and @xmath43 . the temperature of an uncharged , nonrotating black hole is @xcite @xmath44 at all temperatures , a black hole will radiate the massless elementary particles : @xmath45 neutrinos ( if we ignore neutrino mass ) , @xmath46 photons and @xmath47 gravitons . massive elementary particles are evaporated in significant numbers once the peak in the energy distribution given by eq . is of order their particle rest mass . since the exponentially damped tail in eq . extends to infinite energies , a nonzero contribution from any massive species is always present to some extent . at high energies , the sum over @xmath48 and @xmath35 of the absorption probabilities for both massless and massive species approaches the geometric optics limit @xcite : @xmath49 at low energies , @xmath50 is suppressed for massless @xmath51 and @xmath45 species : @xmath52 for @xmath46 and @xmath47 massless bosons , the @xmath53 fall - off is steeper : @xmath54 the absorption probability for nonrelativistic massive particles is more strongly suppressed at low energies than for massless or relativistic particles . for @xmath55 , where @xmath56 is the particle rest mass energy , @xmath57 at @xmath58 differs from the corresponding @xmath59 value by up to 45% @xcite . more importantly , since @xmath38 is the total energy , the spectrum of a massive species must be zero for @xmath60 . these effects have a strong damping influence on the emission spectrum when the peak energy is close to the particle rest mass threshold . prior to particle decays , the peaks in the flux per particle mode measured at infinity occur at @xcite @xmath61 and the total instantaneous flux emitted by a black hole is @xmath62 where @xmath63 is the number of degrees of freedom ( or modes ) per particle species @xmath64 . the standard glashow - weinberg - salam model with three generations has 2 modes per @xmath45 neutrino or antineutrino , 4 modes per @xmath45 @xmath65 , @xmath66 or @xmath67 lepton , 12 modes per @xmath45 quark flavor , 16 modes per @xmath46 gluon , 2 modes per @xmath46 photon , and 2 modes per @xmath47 graviton . with @xmath68 elementary modes for @xmath69 , the total emission rate of particles ( prior to their decays ) is @xmath70 combining eq . with the particle energy gives the instantaneous mass loss of the black hole this can be integrated to derive the lifetime of a black hole of initial mass @xmath71 @xcite : @xmath72 the factor @xmath73 depends on the number of emitted species and is normalized to unity when only massless species are emitted . following the method of ref . @xcite , we can now update the value of @xmath74 , the initial mass of a black hole whose lifetime is the present age of the universe . using the 3-year wilinson microwave anisotropy probe ( wmap ) best - fit value for the age of the universe in the lambda cold dark matter ( lcdm ) model , @xmath75 @xcite , we find that @xmath76 to aid our conceptual discussion of interactions between the emitted particles , we note the magnitudes of various physical quantities . the radius of the black hole is @xmath77 for directly emitted relativistic particles of energy @xmath38 , the reduced de broglie wavelength ( a measure of the effective interaction range for interactions which fall off as some power of @xmath78 ) is @xmath79 in the nonrelativistic case , the factor @xmath80 in eq . is replaced by the momentum . since the flux per @xmath81 or 1 mode peaks at @xmath82 , we have @xmath83 a typical emitted particle therefore has a wavelength comparable to or somewhat larger than the size of the hole @xmath84 . additionally , the particles can be considered ` evaporated ' at a distance @xmath85 since the outgoing solution of the hawking wave equation here already closely approximates the solution for the particle at infinity @xcite . if the time between successive emissions from the black hole is less than @xmath86 , or equivalently if the distance travelled by a particle before the hole emits another particle is less than @xmath87 , we would expect interactions between the particles regardless of their species . equation gives the average time between successive emissions of elementary particles as @xmath88 so @xmath89 and such interactions are unlikely . more precisely , oliensis and hill have calculated numerically that more than 99% of the particles emitted over a hole s lifetime satisfy the condition @xmath90 @xcite . this confirms the appropriateness of using @xmath91 in eq . . similarly , from eq . , once the temperature is above a few times @xmath1 , the energy scale on which the qcd force becomes strong , the time between successive emissions is much shorter than the qcd timescale @xmath92 , and the size of the black hole @xmath84 is much smaller than the associated distance @xmath93 . thus , for @xmath94 , the hawking evaporation of a colored particle is unaffected by the emission of other colored particles and the particle can be regarded as asymptotically free . on the scales associated with @xmath1 in the black hole rest frame , the emission of successive colored particles at these temperatures appears as effectively simultaneous production of highly energetic ( @xmath95 ) quarks and gluons at a point . this is analogous to qcd jet production in @xmath96 annihilation accelerator events . as the quarks and gluons propagate away from the black hole , they should fragment into further quarks and gluons and finally cluster into colorless hadrons once they have travelled a distance @xmath97 from the black hole . here @xmath98 represents an ` average ' energy of the jet particles involved in the decay and @xmath99 is an average ` effective ' rest mass energy . at temperatures above @xmath1 , the observable spectra from the black hole , found by convolving the hawking formula with jet decays , are dominated by the qcd decay products . once the emitted particles have decayed into the astrophysical stable species , the average energy for @xmath100 scales as approximately @xmath101 for @xmath10 , @xmath26 , @xmath27 , @xmath15 and @xmath16 , and @xmath102 for @xmath11 and @xmath103 , rather than as @xmath104 as given by eq . the particle flux after decay , @xmath105 , for @xmath100 scales as approximately @xmath106 @xcite , rather than as @xmath104 as implied by eq . . we first focus on the qed interactions in the heckler scenario , as originally derived in refs . consider a @xmath107 black hole which is emitting hawking radiation in random radial directions . from eq . the number density at radius @xmath108 of electrons and positrons which were directly radiated by the black hole ( and not the result of particle decays ) is @xmath109 where we are now using units with @xmath110 . if the number density is sufficiently high , the electromagnetic particles could lose energy via two - body bremsstrahlung @xmath111 and pair production @xmath112 as they propagate away from the hole . each such interaction would increase the number of electromagnetic particles by a factor of @xmath113 . to calculate the number of scatterings per particle , ref . @xcite takes the bremsstrahlung cross - section in the center - of - mass frame averaged over the photon energy @xmath114 to be @xcite @xmath115 the center - of - mass frame for each interaction is assumed to be the rest frame of the black hole . in eq . , @xmath116 is the energy of the initial electron or positron . since the average momentum exchanged by the virtual photon in the bremsstrahlung interaction is @xmath117 , the heisenberg uncertainty principle implies that the two initial particles must be within @xmath118 of each other to interact significantly . in ref . @xcite the ` formation ' distance for the final on - shell electron and photon is also assumed to be @xmath118 in the center - of - mass frame of the interaction . in the relativistic limit , the bremsstrahlung cross - sections for two incident electrons or two incident positrons or an electron and a positron are the same . the pair - production cross - section can be derived from the bremsstrahlung cross - section using the substitution law ( see section iv.k ) and has the same functional form in the relativistic limit @xcite . in ref . @xcite , @xmath119 in eq . is replaced by an effective electron mass @xmath120 which is taken to be the vacuum electron mass @xmath119 augmented by a ` plasma mass ' @xmath121 to account for coulomb screening by the background of other nearby emitted electromagnetic particles . the electron mass is approximated by the correction applicable for an isotropic thermal plasma @xmath122 which includes the further approximations @xmath123 . this increases the effective electron mass , resulting in a decreased cross - section , within a distance @xmath124 of the black hole . however , there is an inaccuracy here in that the emitted particles do not form an isotropic thermal plasma , as pointed out by heckler @xcite and cline , mostoslavsky , and servant @xcite themselves . the number of scatterings , @xmath125 , experienced by a particle by the time it reaches radius @xmath126 is found by calculating @xmath127 with @xmath128 , the mean - free - path of an individual particle , defined as @xmath129 and @xmath130 the factor of @xmath113 accounts for the increase in the number of electromagnetic particles at each bremsstrahlung interaction , and the relative velocity between the interacting particles is taken to be @xmath131 . only the lower limit in integral is important . as @xmath132 , the number density of the emission decreases as @xmath133 and the scattering cuts off . heckler first neglects the plasma mass correction and takes the lower limit in integral to be @xmath134 . in this case , the integral becomes @xmath135 which exceeds @xmath136 for @xmath137 the analysis is then extended to include the expression for the plasma mass given by eq . and the lower limit in integral is taken to be the radius of the black hole . for @xmath108 small enough that @xmath138 , the number of scatterings is damped by the plasma mass . the dominant contribution to the integral then comes from the value of @xmath108 at which @xmath139 and this gives @xmath140 which exceeds 1 for @xmath141 on numerically integrating eq . , heckler finds that @xmath142 when @xmath143 . thus , although the cross - section @xmath144 remains small , the growth in number density around the black hole at high @xmath104 leads to significant interaction in this model . the region for which @xmath145 is the photosphere region . it is claimed that when @xmath146 the multiple scatterings are sufficient to produce an interacting quasithermal fluid , which flows away from the black hole . the photosphere is found to have inner and outer radii given by @xmath147 approximating this region by a perfect fluid , the average energy of the particles at the outer boundary of the photosphere , in effect the observable average energy of the emitted particles , is estimated to be @xmath148 which is considerably less than the energy @xmath149 expected without photosphere development . ( these values do not include the electrons and photons produced by the decays of other particles radiated by the black hole . ) therefore the flux of photons observed from a black hole with a qed photosphere is @xmath150 neglecting particle decays , compared to @xmath151 without a photosphere . extending the heckler model to qcd effects , eq . is replaced in ref . @xcite by the cross - section for the gluon - bremsstrahlung process @xmath152 : @xmath153 where @xmath22 is now the strong coupling constant . the effective quark mass , @xmath154 , is also enhanced by a plasma term : @xmath155 proceeding as before , one arrives at a rough estimate for the onset of a photosphere due to qcd interactions : @xmath156 when @xmath104 exceeds @xmath157 the precise value of @xmath158 is sensitive to threshold effects and the energy dependence of @xmath22 around @xmath1 . because @xmath104 must be greater than about @xmath1 for a black hole to emit qcd particles , @xmath158 can not be less than about @xmath1 . cline , mostoslavsky , and servant @xcite adapted a code developed to study heavy ion collisions to obtain numerical solutions of the boltzmann equation in the context of heckler s model . as discussed in section iv.l , these authors found that qed and qcd photospheres may form but disagreed with some features of the original heckler model . once the photosphere formed , they also considered the effect of elastic compton scattering ( which conserves particle number ) and found that this can extend the photosphere when @xmath159 . however , because cline , mostoslavsky , and servant employed the same initial description of interactions as heckler , their work is not an independent derivation or confirmation of photosphere development . in this section , we show that several aspects of the original heckler model are invalid and conclude that qed photosphere development is ruled out . we begin by focusing on the qed bremsstrahlung interactions . extension to qed pair production , as well as the corresponding qcd interactions , is straightforward and discussed in section iv.k . as in the original heckler model , we will consider all emitted particles to be initially moving in random radial directions away from the black hole . the effect of black hole size on the emission directions is negligible for @xmath160 . we will also ignore general relativistic effects , which are negligible for @xmath161 . first we consider whether the bremsstrahlung cross - section used by heckler is appropriate . the double - differential cross - section for two - body electron bremsstrahlung is given in ref . @xcite . in this interaction , two electrons and/or positrons exchange a virtual photon , forcing one electron off - shell . the off - shell electron then decays into an on - shell electron by emitting a photon . the eight relevant feynman diagrams for lowest order perturbation theory are shown in fig . [ f1 ] @xcite . the cross - section eq . applies in the relativistic limit for @xmath163 , @xmath164 and @xmath165 bremsstrahlung because the interactions associated with the exchange diagrams , figs . 1 ( e ) , ( f ) , ( g ) and ( h ) , are suppressed at high energy . ( in the nonrelativistic limit , the @xmath165 bremsstrahlung cross - section is greater than the @xmath163 and @xmath164 cross - sections due to the dipole nature of @xmath165 @xcite . ) in the relativistic limit , the average angle between the final on - shell electron and photon is @xcite @xmath166 in the center - of - mass frame of the interaction , where @xmath38 is the energy of each initial electron . the average final energies of the on - shell electron and photon are @xcite @xmath167 because @xmath168 is so small , the majority of interactions involve very small momentum transfer , typically of order @xmath119 . thus the directions of the electrons are modified little by the interaction but their energy loss can be substantial . since the average energy of the final photon is of order @xmath38 , it is valid to use the energy - averaged two - body bremsstrahlung cross - section given by eq . which subsumes the infrared divergence as @xmath169 . in ref . @xcite , @xmath170 is evaluated by numerically integrating the complicated double - differential cross - section . the cross - section does not contain the collinear divergences which would arise at @xmath171 and @xmath172 if factors of order @xmath119 were not correctly included in the derivation . in the non - relativistic limit , @xmath162 decreases more strongly with decreasing @xmath38 than eq . @xcite and becomes process - dependent @xcite , as noted above . it should be stressed that the heckler effect is not due to an infrared or collinear divergence in the cross - section but to the high number density of emitted particles flowing isotropically and relativistically away from the black hole . as a consequence of the emission distribution , the center - of - mass frames for most of the interactions are not moving highly relativistically with respect to the black hole , as we next show . in section ii , we saw that one should not expect qed or qcd interactions between emitted particles if the interaction scale falls off as @xmath78 ( i.e. if the cross - section falls off as @xmath173 ) . in the heckler scenario a quasithermal photosphere develops from qed and qcd bremsstrahlung and pair - production interactions , which do not obey this constraint . we now demonstrate that particles moving in similar radial directions do not interact significantly even via bremsstrahlung or pair production . consider a typical directly emitted particle of rest mass @xmath174 and energy @xmath116 at a distance @xmath160 from the black hole . at this distance , special but not general relativistic effects are relevant . in the rest frame of the black hole , the particle has an individual lorentz factor @xmath175 and velocity @xmath176 to maximize the possible effect , we allow for the small deviations from the radial direction due to the finite size of the black hole . from simple geometry ( see fig . [ f2 ] ) , the particle will be moving at an average angle @xmath177 from the radial direction . note that , for @xmath160 , almost all of the velocity is radial and @xmath178 as @xmath132 . the transverse component of the particle velocity in the black hole rest frame is @xmath179 giving a radial velocity component of @xmath180 the radial velocity is approximately equal to @xmath181 , the velocity of the center - of - mass ( cm ) of two typical particles moving at an angle @xmath182 either side of the radial direction . the corresponding lorentz factor for the cm of the particles is @xmath183 at large @xmath108 , this reduces to @xmath184 , as expected . the special relativistic transformation of velocities is @xmath185 where @xmath186 is the relative velocity of the reference frames , taken to be in the @xmath187 direction . thus the relative velocity of the two particles in their cm frame is @xmath188 where the prime indicates the cm frame . this is close to @xmath136 for radii less than @xmath189 but very small at distances much larger than this . the total number density of emitted particles at a radius @xmath108 in the black hole rest frame is given by eq . . in the cm frame of the two particles , the number density at @xmath108 is therefore @xmath190 hence the number of interactions per particle , integrated over all radii , is @xmath191 taking the cross - section in the cm frame to be the electron bremsstrahlung cross - section ( the pair - production cross - section is similar ) , and allowing for the plasma mass correction ( see section iii ) , we find @xmath192 which gives @xmath193 for @xmath194 . the dominant contribution to the integral comes from the lower limit , here taken to be the radius of the black hole . near this limit , the second term in the first brackets and the first term in the second brackets can be neglected , so we obtain @xmath195 hence emitted particles traveling in a similar radial direction will not interact with each other via bremsstrahlung or pair production , or scatter or lose energy to any significant degree , as they propagate away from the black hole . this is a consequence of the lorentz factor of the emitted particles being so large in the black hole frame . we now examine the heckler assumption that the cm frame for most interactions is the rest frame of the black hole , so that @xmath162 is given by eq . in the black hole frame . as we have seen , this is not true for particles moving in similar radial directions . here we investigate particles moving in different radial directions . consider ultra - relativistic particles 1 and 2 of the same energy and rest mass , created near the black hole at the same time and subsequently moving radially outwards with speed @xmath196 along rays separated by an angle @xmath197 . the lorentz factor @xmath198 of the cm of the two particles relative to the rest frame of the black hole is @xmath199 this is approximately independent of @xmath200 ( and hence @xmath104 ) when @xmath201 . equation implies that @xmath202 when @xmath203 ; @xmath204 when @xmath205 ; @xmath206 when @xmath207 ; and @xmath208 when @xmath209 . thus the cm frames for the interactions of particle 1 with the majority of particles streaming out from the black hole are not significantly lorentz - boosted relative to the black hole . the region for which the interaction frames are significantly lorentz - boosted relative to the black hole is a cone of angle @xmath210 around the direction of particle 1 . in general , the energies of particles 1 and 2 are not equal but have the hawking distribution , given by eq . , and the corresponding formula for @xmath198 is more complicated . let us denote the ultra - relativistic speeds of particles 1 and 2 by @xmath211 and @xmath212 , respectively , where @xmath213 , and let @xmath214 be the angle between @xmath215 and the velocity of the cm of particles 1 and 2 , @xmath216 . the special relativistic equations for relative velocity then give @xmath217 and @xmath218 for @xmath219 . again @xmath220 is approximately independent of @xmath200 and @xmath221 ( and thus @xmath104 ) unless @xmath197 is extremely close to zero . if we assume @xmath222 , without loss of generality , and consider @xmath223 , corresponding to @xmath224 ( this more than adequately covers the majority of the hawking distribution around @xmath91 ) , we have @xmath225 when @xmath226 and @xmath227 when @xmath228 . similarly if @xmath229 , then @xmath225 when @xmath230 and @xmath227 when @xmath231 . hence we conclude that taking the cm frame of each interaction to be the black hole rest frame , as heckler does , is a valid approximation for the interactions of any given particle with most other particles . to within a factor of @xmath232 , using @xmath162 in the black hole frame to describe the interactions of the particle is correct , except for an exclusion cone of angle @xmath233 radians around the direction of the particle s velocity for which the interactions are negligible because @xmath234 . however , as we shall see below , this exclusion cone has important consequences which hinder the development of the photosphere . once the particle has traveled a distance @xmath235 from the black hole , the transverse distance between the particle and the edge of its exclusion cone is @xmath236 because @xmath237 is a transverse distance , @xmath237 is the same in the particle and black hole frames . thus once the particle has traveled a distance @xmath238 from the black hole , the distance to the nearest particle with which it could then interact is of order @xmath134 , provided the emitted particles are still traveling mainly radially . because @xmath162 dictates that two particles must be within @xmath118 of each other to interact via bremsstrahlung , few interactions can occur after the particles have propagated this distance . the relevant interactions for initiating photosphere formation can only be those which occur within @xmath239 of the black hole . this point is not included in the original heckler model . as we shall demonstrate in section iv.g , the electron coming out of a bremsstrahlung interaction is still off - shell at @xmath239 . also @xmath162 is truncated by the causality constraint and may be damped by off - shell ( lpm - type ) interactions , as we discuss in sections iv.f and iv.h . these effects significantly decrease the capacity for photosphere formation . we have assumed in the exclusion cone argument that the emitted particles move out radially from the black hole we now justify this . were they to random walk out , an individual particle would have to undergo @xmath240 scatterings to deviate by an angle @xmath233 from the radial direction , where @xmath168 is given by eq . . if @xmath241 gev , @xmath242 scatterings are required for the particle to scatter outside the exclusion cone . thus even if a photosphere can develop , particles will deviate little from the radial direction before and after photosphere formation . ( as discussed in section iv.l , this is incorrectly treated in the numerical work of cline , mostoslavsky , and servant @xcite . ) therefore the constraint of section iv.d set by the transverse distance to the edge of the exclusion cone will also apply after the onset of scattering . this confirms that the only relevant interactions can be those which occur within @xmath239 of the hole . additionally , if we consider the possible deviation from the radial direction due to the finite size of the black hole , eq . implies that @xmath243 only when @xmath244 . this is approximately the radius at which the hawking emitted particle first appears and , for @xmath107 , represents a tiny portion of the region within @xmath239 of the black hole . hence the effect of black hole size on the propagation direction should also be negligible . we now calculate the number of interactions @xmath125 each particle undergoes . the definition of the mean - free - path used in the heckler model and given by eq . is clearly wrong for geometrical reasons . equation applies when a particle propagates linearly through a target of uniform number density @xmath48 in the configuration shown in fig . [ f3 ] @xcite . this is not geometrically analogous to the black hole situation for three reasons . first , the particles are moving radially outward in a spherical distribution from the black hole . second , the number density @xmath245 varies with the distance of the moving particles from the black hole . that is , if we consider particle 1 interacting with an element of the flux around particle 2 , @xmath246 does not remain constant as particles 1 and 2 travel outwards . third , implicit in eq . is the assumption that the interacting particles have an infinite past and future , which is not the case for black hole emission . consider particle 1 at radial position @xmath247 and particle 2 at @xmath248 with @xmath249 . particle 2 was emitted by the black hole at a time @xmath250 later than particle 1 . the hawking derivation of eq . utilizes the solutions at infinity but does not tell us how a particle is evaporated : either the particle did not exist or it was hidden by the event horizon before emission . in either interpretation , the particle prior to its emission can not interact with particles outside the event horizon . the interaction lifetime of the particle is therefore truncated . because @xmath125 in eqs . and is dominated by the interactions of particle 1 with the short - lived , high density particles close to the black hole , this truncated lifetime should have a significant effect on the heckler scenario . an important consequence of the truncated lifetime is its implication for causality . from the classical point of view , i.e. ignoring temporarily the quantum smearing of particle space - time position and energy - momentum due to the heisenberg uncertainty principle and the quantum nature of interactions , two particles can interact only if they are causally connected or , equivalently , if a light signal can be transmitted between them . classically each particle is point - like and emitted by the black hole at a precise time . without loss of generality , let us assume particle 1 , emitted at time @xmath251 in the black hole rest frame , is the scattering particle in the bremsstrahlung interaction and particle 2 , emitted at @xmath252 , is the scattered particle . the time difference @xmath253 may be positive or negative . the most appropriate frame for analyzing the causality constraint is the frame of the scattering particle because the coulomb field of the scattering particle has an undistorted @xmath254 distribution in its own frame . let us consider a light signal which travels a distance @xmath255 from the scattering particle in the scattering particle s frame and reaches the scattered particle . the form of the invariant @xmath162 , eq . , implies that , if the particle histories were not truncated , the probability for the scattered particle to undergo a bremsstrahlung event would be @xmath256 if @xmath257 , but negligible if @xmath258 . ( the truncation of the particle histories will further suppress the probability within @xmath257 @xcite . ) if @xmath259 and particle 1 emits a light signal which reaches particle 2 just as particle 2 is emitted by the black hole , one can readily calculate @xcite that @xmath260 where @xmath261 is the velocity of the particles in the black hole frame . if , alternatively , @xmath262 and particle 1 , as soon as it is evaporated by the black hole , emits a light signal which then reaches particle 2 , a slightly longer calculation @xcite shows that @xmath263{\rm \ ; } \left\|\delta t\right\|>\gamma v\left\|\delta t\right\| \label{52}\ ] ] where @xmath197 is the angle between the particles in the black hole frame . thus in either case , for the @xmath264 constraint to be met , the difference in the emission times must be @xmath265 for relativistic particles . however , from section ii , the average time between successive emissions of electrons and positrons directly evaporated by the black hole is @xmath266 using @xmath267 , we see that @xmath268 for all @xmath194 . thus the classical causality condition is only satisfied by a negligible fraction of the emitted particles when @xmath194 . the same conclusion applies for qcd particles when @xmath269 . because the emission of a given species is damped near its rest mass threshold @xcite , so increasing @xmath270 , the classical causality condition is unlikely to be satisfied even near particle rest mass thresholds . ( also recall that @xmath162 decreases faster with decreasing @xmath38 in the non - relativistic limit . ) at any @xmath104 , if a photosphere is to form , we additionally require the causality constraint to be met within a sphere of radius @xmath238 around the black hole . this only further tightens the @xmath253 constraint . quantum mechanical effects relax the above classical constraint only slightly . by the heisenberg uncertainty principle , the emission of each electron from the black hole is smeared out over a time @xmath271 , leading to it having a spatial spread in the direction of emission of @xmath272 . thus when the electron is a distance @xmath108 from the black hole , we can visualize it quantum mechanically as an eggshell segment of thickness @xmath273 , radial curvature @xmath108 and tangential width growing to @xmath134 , whose probability of overlapping another emitted electron is essentially given by the classical constraint above . additionally , the virtual photon exchanged between the two bremsstrahlung - interacting electrons , which typically transfers 3-momentum @xmath274 and negligible energy and has a nonzero mass ( @xmath275 ) , can not be localized within a length less than @xmath276 . it can extend over a spacelike interval but spacelike intervals in the propagator are strongly suppressed . however , the wave speed at which the field of one electron propagates to the other electron in the dirac equation remains @xmath131 . therefore the above classical causality constraint essentially always applies to the quantum interaction . equation then must be replaced with an expression which more accurately describes the geometry of the interaction and accounts for the shortened histories of the interacting particles . as we discuss in ref . @xcite , the causality constraint and truncated histories also decrease the momentum transferred in each interaction , decreasing the effective @xmath162 well below the already small @xmath144 cross - section of eq . . even at high @xmath104 , the average number of interactions per particle is bounded by @xmath277 and can not reach @xmath142 . if an emitted particle does scatter once , we now show that the distance to complete that first scattering is such that subsequent complete scatterings by the particle within @xmath239 of the black hole are not possible . thus it is improbable for particles to experience more than one bremsstrahlung interaction . this by itself is sufficient to prevent qed photosphere formation . in the original heckler model the distance needed for the completion of a bremsstrahlung interaction , i.e. the distance required for the formation of the final on - shell electron , is assumed to be @xmath278 in the cm frame of the interaction . this is not correct and must be replaced by the much larger distance @xmath279 , as we can see in two ways . first , the 4-momentum of the off - shell electron is @xmath280 where @xmath281 and @xmath282 are the 4-momenta of the final on - shell electron and photon , respectively ( see fig . 1 ) . from eqs . and , the average energies of the final electron and photon are @xmath283 and the average angle between them is @xmath284 , so we have @xmath285 . thus the average lifetime of the off - shell electron in its own frame is @xmath118 , which corresponds to @xmath0 in the cm frame of the interaction . second , following ref . @xcite , consider the final electron and photon as they are being created . because they are created with a small separation angle , their wave packets will initially overlap for some distance as they propagate . this distance is the formation distance . the transverse momentum components of the final electron and photon relative to the momentum of the initial electron @xmath286 are both @xmath287 . by the heisenberg uncertainty principle these transverse momenta can be resolved only when the transverse spatial separation is at least @xmath288 . since the opening angle between them is @xmath168 , this occurs after they have traveled a radial distance @xmath289 in the cm frame of the interaction . thus again the formation distance must be @xmath279 . as we have already noted , the consequence of this is that the only relevant interactions for photosphere creation can be those which occur within @xmath238 of the black hole . certainly @xmath290 for @xmath107 and so , on completion of the first bremsstrahlung interaction , the products can no longer significantly interact with other particles . in the extremely rare event that an electron undergoes more than one scattering within @xmath239 of the black hole , the subsequent scattering occurs when the electron is off - shell , not on - shell . in the general setting , if an off - shell particle undergoes multiple scattering over the distance @xmath291 ( the formation length of the photon and final electron in the bremsstrahlung interaction ) , landau , pomeranchuk and migdal have shown that the bremsstrahlung cross - section is decreased by a numerical factor of up to @xmath232 ( see @xcite ; for a recent review see ref . this is known as the lpm effect and has been confirmed experimentally , at least in the soft photon regime @xcite . ( the effect is greatest in this regime because @xmath291 increases as @xmath114 decreases @xcite . ) the suppression of the bremsstrahlung cross - section is due to multiple scattering breaking the coherence of the final state as the photon is forming @xcite . the suppression increases with electron energy and the number of off - shell scatterings or , equivalently , the number density . the lpm effect modifies both qed and qcd bremsstrahlung . recent theoretical treatments include elastic ( coulomb ) and inelastic ( bremsstrahlung ) scatterings of the off - shell electron @xcite . applying the lpm effect to the black hole scenario , we would expect multiple scatterings of the off - shell electron to decrease the effective bremsstrahlung cross - section . since the off - shell electron requires a distance @xmath291 to decay into the final on - shell electron and photon , even if @xmath292 within @xmath238 , each particle generally undergoes at most one complete on - shell scattering as it streams away from the hole . recall that the probability of undergoing even one complete scattering is significantly damped by the causality constraint at @xmath107 . incorporating the precise effect in the black hole scenario requires calculating the lpm effect from first principles because the black hole scenario differs from other lpm scenarios in at least three ways : ( i ) the black hole off - shell interactions are bremsstrahlung ( not just coulomb scatterings ) with average momentum transfer similar to that of the initial bremsstrahlung interaction ; ( ii ) the off - shell interactions and the initial interaction occur essentially simultaneously , not sequentially , and their range is greater than the distance between each off - shell interaction ; and ( iii ) the non - uniform number density , truncated particle histories and energy distributions of the emitted particles are specific to the black hole case . proceeding qualitatively , we expect the following outcomes for black hole qed interactions . for bremsstrahlung , we have @xmath293 , so multiple scatterings , if they occur , can perturb only a small segment of the complete interaction of a typical particle . because @xmath294 , any scatterings affect the particle essentially simultaneously within a distance @xmath295 of the black hole . ( the actual region will be smaller than @xmath239 because of the causality constraint . ) therefore the particle can be thought of as interacting with a region within @xmath295 as a whole . this corresponds to the single radiator or thin target lpm limit @xcite . the probability for interaction in this limit is given by the total scattering angle within the @xmath295 region @xcite . if we approximate this angle by that appropriate for a random walk of successive scatterings , @xmath296 where @xmath297 is the number of scatterings the particle undergoes within @xmath295 of the black hole , then @xmath298 for @xmath299 . from the numerical simulations of haug @xcite , @xmath300 falls off roughly exponentially at angles above @xmath168 and the momentum transfer is increased , although the energies of the final photon and electron stay of order @xmath38 . hence @xmath301 and the modified cross - section is always less than that for the interaction without off - shell scattering . additionally , if the bremsstrahlung photon pair - creates before it is completely formed , @xmath162 is further suppressed @xcite . thus the net effect of multiple off - shell qed scatterings is to decrease the probability of any particle producing a bremsstrahlung photon . even if @xmath292 , each particle requires a distance @xmath294 to complete the interaction and so is unlikely to undergo more than one completed bremsstrahlung interaction as it streams away from the black hole . because each particle rarely undergoes more than one completed interaction , the number density around the black hole can increase by only a small fraction ( at most 50% ) even if @xmath292 . hence we conclude that a qed photosphere will not develop . in the original heckler model , it was argued that on scales less than @xmath238 the qed bremsstrahlung cross - section should be suppressed by coulomb scattering off other electrons around the black hole . ( however , heckler did not include the off - shell bremsstrahlung interactions discussed above . ) to account for the coulomb scattering , refs . @xcite took the simplified approach of defining the effective electron mass @xmath119 to include the fermion self - energy for a finite temperature bath . in the expression for @xmath162 , @xmath119 was augmented in the interaction rest frame by the thermal plasma mass given by eq . , which describes the screening of the electron by other emitted electrons . as heckler pointed out , the hawking emission is not a true plasma . because the emitted particles flow isotropically out of the point - like black hole , the particle distribution can not be transformed into a frame in which the particles exhibit a maxwell - boltzmann momentum distribution . as we have seen , the causality constraint prevents emitted electrons interacting significantly within @xmath238 of the black hole . this should apply to coulomb scattering as well as bremsstrahlung interactions . also , since each electron requires a distance @xmath294 to complete a bremsstrahlung interaction , @xmath245 can not increase substantially within @xmath239 of the black hole due to interactions . thus plasma mass considerations should be irrelevant within @xmath238 for @xmath107 . although the heckler photosphere scenario relies on the @xmath302 two - body bremsstrahlung interaction between two charged particles emitted by the black hole , the emitted particles may experience other qed bremsstrahlung processes which we investigate in detail in the accompanying paper @xcite . first , there should be the @xmath303 inner bremsstrahlung generated by the charged particle as it escapes the black hole and , from the point of view of a distant observer , has its velocity changed from being essentially zero at the black hole to its asymptotic value . this inner bremsstrahlung emission will contribute about 10% of the total photon power from an @xmath304 black hole and has an approximately energy - independent power spectrum which dominates over the directly emitted hawking photons at low energies @xcite . secondly , there should be the @xmath305 bremsstrahlung emission from the scattering of the charged emitted particles by the stochastic electromagnetic field of the black hole , and another @xmath302 contribution from the scattering of the inner bremsstrahlung photon off an ambient charged particle . in fact , since the inner bremsstrahlung dominates all bremsstrahlung processes , the other bremsstrahlung processes can be simply regarded as small @xmath303 and @xmath305 corrections to the inner bremsstrahlung . the total bremsstrahlung emission is then essentially determined by the asymptotic momentum distribution of the charged particles , which to lowest order in @xmath306 is determined by the hawking emission formula eq . and to higher orders is modified slightly by the interactions between the emitted charged particles . in all cases , however , the outgoing photon and electron from any bremsstrahlung interaction require a formation distance @xmath307 . thus none of these bremsstrahlung interactions can produce a qed photosphere around a @xmath107 black hole . so far , we have discussed qed bremsstrahlung interactions but the above considerations also apply to qed pair production @xmath112 . the pair - production cross - section is obtained from @xmath162 by rotating the feynman diagrams to depict an incident photon rather than an outgoing photon and using the substitution rules @xcite . in the relativistic limit , @xmath308 has the same form as @xmath162 @xcite . the distribution of the outgoing @xmath26 and @xmath27 is strongly peaked in the direction of the initial photon and each carries on average approximately half the initial photon energy @xcite . the formation length of the @xmath309 pair is also @xmath310 when @xmath311 @xcite . as before , any multiple off - shell scattering will suppress @xmath308 within the distance @xmath291 when @xmath107 . thus each particle can undergo at most one complete pair - production event as it streams away from the black hole and the probability of that interaction is strongly suppressed by the causality constraint . again no qed photosphere can develop . the above discussion also applies to qcd gluon - bremsstrahlung and qcd pair - production interactions of the emitted particles . ( note here we are referring to the qcd interactions between particles in different qcd jets , not between particles in the same qcd jet . the interactions between particles in the same jet are included in the fragmentation and hadronization modeling of ref . @xcite . ) to derive the corresponding qcd quantities , @xmath119 is replaced by the quark mass @xmath154 or @xmath1 , and @xmath306 by the strong coupling constant @xmath22 . this leads to eq . in the relativistic case , neglecting color factors of @xmath232 . the average time between qcd particle emissions by the black hole is @xmath312 when @xmath94 . the qcd formation scale @xmath313 is also the scale at which the hawking emission hadronizes . applying the same arguments as above , when @xmath94 , the causality constraint implies that the emitted qcd particles have little chance of interacting within @xmath314 of the black hole and any particles that do interact can undergo at most one complete qcd bremsstrahlung interaction within @xmath315 . thus no qcd photosphere ( also known as a chromosphere ) should form when @xmath94 . because of the time separation between subsequent qcd jets and the lpm effect , nor should a photosphere form from interaction of the hadronization products which appear at @xmath316 . qcd phenomenology and experiments also strongly argue against a qcd photosphere developing when @xmath104 ( or more specifically the peak energy of the directly emitted flux ) passes through values that are roughly @xmath1 . because the heckler estimate , @xmath317 , is close enough to @xmath1 that high energy formulae may be inappropriate , the question arises of whether a qcd photosphere could form from _ non - relativistic _ effects at these temperatures . if the emission were non - relativistic , the causality constraint would be weakened . also at low energies the effective @xmath22 is observed to increase from its high energy limit , @xmath318 where @xmath319 is the number of relevant quark species , and may approach @xmath320 as @xmath321 @xcite . on the other hand , it is not clear whether , even if multiple scatterings were to occur around @xmath322 , they would affect the observable spectra , because the hawking emission is strongly damped and the multiplicity per qcd jet is very low around the @xmath1 threshold @xcite . however , we deduce from both the qcd phenomenology and experimental perspectives that when the black hole begins to emit asymptotically free quarks , these quarks are relativistic . therefore no qcd photosphere should develop around @xmath322 . the emission picture which is consistent with qcd phenomenology and threshold accelerator experiments is as follows . when @xmath104 ( or more precisely @xmath323 for @xmath51 ) is about the pion rest mass ( @xmath324 and @xmath325 ) , the black hole should start emitting pions directly , with the grey - body factors and degrees of freedom appropriate for massive @xmath51 particles modified by the finite - size structure of the pion . once the scale of the black hole emitting region becomes small compared with the length below which the constituent ( valence ) @xmath326 and @xmath235 quarks in the pions become asymptotically free , the black hole should begin emitting the quarks individually . these quarks then hadronize into pions on the proper distance scale @xmath327 in the frame of the emitted quark . the energy at which this happens is sufficiently above the pion rest mass that the pion is relativistic . hence the pion s constituent quarks should also be regarded as relativistic . furthermore , during the asymptotic freedom regime before the quark hadronizes , the quark masses which are used in the qcd lagrangian are the much lighter quark current masses , not the quark constituent masses . ( the quark current masses are @xmath328 , @xmath329 and @xmath330 , whereas the quark constituent masses are @xmath331 and @xmath332 . for the heavy @xmath131 , @xmath333 and @xmath334 quarks , the current masses are approximately equal to the constituent masses and lie above @xmath335 @xcite . the quark constituent masses are the relevant masses as the quarks hadronize and afterward . ) thus the quarks are ultra - relativistic during their asymptotic freedom phase and then gain larger effective masses , but still remain relativistic , when they hadronize . the hadronization process itself at these energies also can not produce a significant increase in the number of final particle states . what the quarks hadronize into is dictated by energy and quantum conservation laws . hence , around @xmath322 , the quarks can hadronize only into pions ( or the few light meson states which decay into pions or mimic @xmath336 decay into photons ) and the multiplicity of the final pions per initial hawking emitted quark must be very low ( one or two pions per initial quark ) @xcite . emission of @xmath309 or photons by these quarks or pions is covered by our earlier analysis of qed interactions and so is negligible , as are any weak interactions . gluon bremsstrahlung by quarks at these energies is also inconsequential . the gluon has to obey the qcd confinement ( no - free - color ) and quantum conservation rules . thus a gluon must have an energy of about @xmath1 or greater to produce free final state particles , the lowest energy option being a pion pair . because the total momentum in the frame defined by the black hole and the hawking emitted quark must be conserved and there are no available very light final states for the gluon or the quark , an asymptotically free hawking emitted quark of energy @xmath337 can not be slowed down by gluon bremsstrahlung , except by the gluons involved when that quark eventually hadronizes . the above description is completely consistent with accelerator experiments . the cross - section for @xmath338 has been measured at center - of - mass energies from close to the @xmath339 threshold to 1 gev and remains smooth over that region ( see ref . @xcite for a recent review ) . no ` dip ' or other structure in the data is observed around center - of - mass energies of @xmath1 where the behavior changes from direct production of pions to the production of final state pions via an intermediate asymptotically free quark state . to further visualize the black hole quark emission and hadronization process around @xmath322 and at higher temperatures , one can use the chromoelectric flux - tube model ( also known as the qcd string model ) that successfully describes hadronization in accelerators ( see @xcite , @xcite ; for recent reviews see refs . @xcite ) . in this interpretation a quark is emitted from the black hole dragging behind it a tube of color field whose other end is attached to the black hole . the string tension @xmath340 of the flux - tube is constant , so that its potential energy @xmath341 grows linearly with the length of the tube @xmath187 . when the flux - tube has been stretched to the length @xmath342 at which the potential energy matches the energy required to create a new @xmath343 pair out of the vacuum ( including the transverse momentum of the @xmath343 pair implied by the heisenberg uncertainty principle applied to the flux - tube width ) , the tube breaks in two , producing a @xmath36 and a @xmath344 attached to the two new ends where the tube broke @xcite . to conserve energy , the stretching and @xmath343 creation is accompanied by a corresponding deceleration of the lead quark this stretching and breaking sequence continues until the energy of the lead quark has been expended producing new @xmath343 pairs . the quarks and antiquarks created from the flux - tube match up to hadronize into bound color - neutral states . the whole process is completed over the distance scale @xmath345 in the black hole frame , where @xmath38 is the energy of the initial quark . the flux - tube model naturally gives a multiplicity growth of final states per initial quark that is proportional to @xmath346 , as observed in accelerators . that the asymptotically free quarks are relativistic , rather than non - relativistic , when first emitted by the black hole around @xmath322 can additionally be seen from the thermodynamical ` detailed balance ' derivation of the hawking flux . when a black hole is in quasithermal equilibrium at a given instant with a surrounding thermal bath of the same temperature , the absorption rate of a given species equals its emission rate . if a pion incident upon a black hole has total energy insufficient for the pion to break up into subparticles with one subparticle going down the hole and one escaping , detailed balance gives the hawking emission rate of direct outgoing pions from the capture cross - section of the incoming pions . when the pion size is comparable to the cross - section for a point - like scalar particle of the same mass and energy as the pion , the pion capture cross - section should be similar to that of the corresponding point - like scalar particle . ( this cross - section should also apply when the black hole itself is much smaller than the pion if the pion is moving sufficiently slowly that the capture cross - section is much larger than the hole itself . in this case , once the pion falls to a distance from the hole comparable to its size , it will be nearly inevitable that part of the pion will fall into the hole , with the rest having insufficient energy to escape even though the pion may initially appear to have too large an intrinsic size to be captured by the smaller hole . ) if the black hole size is significantly smaller than the size of the pion and the incoming pion is moving relativistically ( that is , its kinetic energy is significantly greater than its rest mass ) , the black hole can capture an individual constituent quark of the pion , with the other constituent escaping capture or being captured after a significant time interval . this inelastic scattering of the pion by the black hole could also lead to the production of other particles , depending on the available energy . one could visualize the capture process as the glue string between the captured quark and the constituent antiquark of the original pion breaking in two , with a new antiquark forming on the end of the string attached to the quark that fell into the hole and a new quark forming at the end of the string attached to the uncaptured antiquark . the new antiquark attached to the glue string going into the hole could be captured by the hole , while the new quark attached to the external glue string could escape along with the original antiquark . for incident pions of high enough energy , more complicated inelastic processes could lead to greater multiplicity and number of species in the escaping and captured states . the reverse process of the absorption of constituent quarks corresponds to the hawking emission of individual relativistic quarks that hadronize into one or more pions and/or other mesons and baryons . from our above discussion it is clear that the qcd production around a @xmath322 black hole is not analogous to the creation of quark - gluon plasma at the relativistic heavy ion collider ( rhic ) ( for a recent review of rhic results , see ref . @xcite ) . at rhic the collision of heavy @xmath347 nuclei at center - of - mass energies of @xmath348 and @xmath349 per nucleon generates a strongly coupled , high entropy quark - gluon plasma which rapidly achieves a thermal spectrum and then expands isentropically until the particles hadronize and decouple by a time @xmath350 . presently it is not known whether the plasma achieves its thermal spectrum of states by interactions or is initially created with that distribution of states @xcite . at the moment of collision , the overlapping @xmath351 nuclei are highly lorentz - contracted in the center - of - momentum frame , implying an overlap time of @xmath352 and an initial thermalized energy density of about @xmath353 . the initial energy density estimate varies with the interpretative model but in all cases is far greater than the @xmath354 threshold for quark - gluon plasma formation or the density within a nucleon @xmath355 , and remains so for up to @xmath356 . the number of charged particles arriving at the detectors ( about @xmath357 ) indicates that approximately @xmath358 high energy particles are created per initial nucleon but the dependence of the charged particle multiplicity on the available energy per colliding nucleon still matches that seen in @xmath359 and @xmath360 accelerator collisions @xcite . the hadronic multiplicities are determined by the nonzero baryonic chemical potential ( baryon / anti - baryon asymmetry ) of the initial nuclei . notably , a reduction of a factor of about @xmath361 in the number of high transverse momentum hadrons is observed . this phenomenon is known as jet - quenching , although the high transverse momentum jets are not truly quenched but spread over a wider angle with lower average energy per particle . the hadronic energy loss implies significant qcd interaction with the ambient plasma , but photons are observed to travel through the plasma unimpeded . the jet - quenching can be modeled by assuming that the collision initially creates a gluon number distribution of @xmath362 , where @xmath363 is the rapidity defined by @xmath364 for a particle of energy @xmath365 and transverse momentum @xmath366 . ( integrating over @xmath367 , this gives the number of gluons per nucleon to be of order @xmath368 ) . application of the heisenberg uncertainty principle indicates that the nuclei are gluon - saturated on the scale of the collision even before impact : that is , at 200 gev per nucleon the two gold nuclei see each other as dense objects , not ensembles of distinct partons , because the atomic number ( parton density ) and transverse momenta are such that the partons overlap considerably in the transverse plane @xcite . all of the above effects attributable to a strongly - coupled quark - gluon plasma are experimentally confirmed to be suppressed at lower collision energies and in deuteron - gold collisions @xcite . in the @xmath322 black hole case , however , in contrast to rhic , the conditions for the simultaneous production of an ultra - high density of qcd particles are not met . a @xmath322 black hole emits particles by the hawking process sequentially and relativistically with significant time separation between each particle emission . thus , as we have seen in section iv.f , negligibly few of the hawking emitted quarks are in sufficient causal contact to interact with each other within @xmath327 of the black hole . furthermore , as discussed above , conservation of energy severely limits the number of final states that can be created per initial quark around @xmath322 . in the black hole case , only an energy @xmath337 is available per hawking emitted quark , in contrast to the @xmath349 per nucleon available at rhic . ( additional differences include the facts that the corresponding baryonic chemical potential is @xmath369 , because the black hole emits equal numbers of quarks and antiquarks , and that the black hole particles are not initially gluon - saturated . ) hence , as given by eqs . , the greatest achievable density around a @xmath322 black hole is well below @xmath354 . the closest experimental analogy to the @xmath322 black hole situation is the generation of pions and qcd jets in @xmath359 accelerator collisions and in hadronic collisions at much lower center - of - mass energies per nucleon . in summary we conclude that the time separation between emissions and the limited average energy available per hawking emitted particle prevent the formation of a qcd photosphere around a @xmath322 black hole , and the causality and @xmath315 constraints prevent qcd photosphere formation at @xmath94 . we also note that the above description of the qcd threshold emission processes should lead to more precise numerical modeling of the astrophysical particle spectra produced by hawking radiation around the @xmath1 threshold than has previously been published . the original heckler scenario was numerically modelled by j. cline , mostoslavsky , and servant @xcite , who simulated photosphere development using the test particle method to solve the boltzmann equation for propagation of plasma particles undergoing collisions @xcite . they found that qed photosphere production set in at @xmath370 ( comparable to heckler s value of @xmath371 ) but extended over a much smaller region : it started at @xmath372 ( compared to heckler s @xmath373 ) and ended at @xmath374 where the trajectories of particles within the interaction range rapidly become parallel , as we saw in section iv.d . for @xmath375 , their photosphere region was extended by compton scattering effects . however , cline , mostoslavsky , and servant use @xmath376 , whereas the correct relativistic klein - nishina cross - section @xcite for a photon of energy @xmath38 compton scattering off an electron of energy @xmath38 in the black hole rest frame is @xmath377 . they also found that the fluid description is of questionable validity and that it led to an average photon energy at the outer photosphere edge which differed from heckler s value , even decreasing as @xmath104 increases . the differences are more marked in the qcd case , partly because of the form of the qcd plasma mass correction used by heckler . they found that a qcd photosphere forms for @xmath378 and extends to @xmath379 . the cline , mostoslavsky , and servant work is elucidating but , because it takes the same approach as ref . @xcite , all of the points we have made above apply to it . most importantly , the authors neglect the causality and @xmath291 constraints which limit each particle to at most one on - shell bremsstrahlung interaction within @xmath374 or @xmath379 and make even that highly unlikely . an additional problem is that they generate the bremsstrahlung photons with random energy @xmath380 and random angle @xmath381 . we have seen that the true photon angle is very small @xmath382 and deviates little from the direction of the initial electron . randomizing the angle would exaggerate photosphere development , and this may explain why cline , mostoslavsky , and servant find @xmath24 to be lower than heckler s original analytic estimate . their numerical method should therefore be modified to include all of our previous points in section iv and could be employed to investigate the consequences of the rare bremsstrahlung interactions . kapusta @xcite and daghigh and kapusta @xcite have analytically and numerically investigated the hydrodynamic expansion of matter around a black hole in the original heckler photosphere formation scenario . in their treatment , the @xmath383 emission is assumed to be a quasi - stationary shell of plasma expanding relativistically away from the hole . the plasma is kept in local thermal equilibrium by the qed and qcd bremsstrahlung and pair - production interactions of the original heckler model . ` local thermal equilibrium ' in this context is taken to mean that the mean - free - path , or thermalizing scale , of the plasma particles is less than the scale over which the plasma temperature is changing @xcite . they find that the quasi - stationary photosphere shell extends out to @xmath384 . the expected observable photon signal is then calculated from the luminosity of the photosphere surface . the results of daghigh and kapusta are not supported by our analysis . as we have seen in section iv , the interactions within @xmath295 of the black hole can not significantly increase the number density @xmath245 , as required for photosphere formation . their approach also neglects the causality and @xmath291 constraints within @xmath295 . once a particle reaches @xmath239 , we have shown that the geometry is such that it has little probability of interacting with other emitted particles . thus , contrary to the claim of daghigh and kapusta , there should be no change in the particle energy after the particle reaches @xmath385 and no persistent qed or qcd photosphere or locally thermalized plasma should develop . in section iv we have shown that a heckler qed photosphere will not form around a hawking black hole . similar arguments and additional qcd phenomenological considerations prevent a qcd photosphere developing . interactions which occur within @xmath239 of the black hole determine whether a heckler photosphere can arise . these interactions are extremely rare and any multiple scatterings involve off - shell particles . in contrast , the original heckler model postulated that each particle on average undergoes successive on - shell bremsstrahlung interactions out to a much greater radius . this led to an increase in the expected flux at low energies and a steepening of the spectra at high energies . including the off - shell scatterings , however , suppresses the effective bremsstrahlung cross - section and permits at most one completed bremsstrahlung interaction per particle . this one net interaction can only slightly modify the observable emission spectra , and not significantly as would a photosphere . our remarks apply equally to pair production . also of crucial importance to determining the full spectra is understanding the emission processes around @xmath322 . if each emitted particle produces at most one bremsstrahlung particle or undergoes pair - production once as it streams away from the black hole , there will be little effect on the detectability of an individual black hole or a background of black holes . the number of relativistically emitted particles of the relevant species in the relevant part of the spectra could be increased by at most a factor of @xmath113 and the average energy decreased by at most @xmath386 . furthermore , given the low likelihood of even one net interaction , the causality constraint discussed in section iv.f implies that any change to the spectra should be several orders of magnitude weaker than this . as we show in ref . @xcite , inclusion of the causality constraint in an impact - parameter - based analysis of the two - body bremsstrahlung interaction around a @xmath107 black hole leads to an upper limit of @xmath387 on the total fraction of energy emitted in two - body bremsstrahlung photons . charged and/or colored particles ( @xmath65 , @xmath66 , @xmath67 , @xmath388 , @xmath389 ) and photons which are directly emitted by the black hole could participate in the rare bremsstrahlung or pair - production interactions . additionally , almost all of the @xmath10 , @xmath11 , @xmath103 , @xmath26 , @xmath27 , @xmath15 and @xmath16 flux from a @xmath390 black hole comes from the decays of directly emitted particles . thus all observable spectra , except the highest energy @xmath391 flux which is directly emitted , may be very slightly modified by interactions . as we discuss in ref . @xcite , at low energies the photon spectrum from an individual black hole will be enhanced by two other @xmath303 and @xmath305 bremsstrahlung processes . in contrast , the low energy particles in the spectra from a pbh background were originally emitted with higher energies and then cosmologically redshifted to low energies @xcite , and so the background spectra can be only slightly modified at low energies ( by the rare interactions near the black hole ) . in summary , the observable spectra from an individual black hole or pbh background remain those of the standard hawking picture . the time - integrated background spectra above @xmath392 mev should maintain their original @xmath393 form @xcite with little amplitude change , in contrast to the @xmath394 slope predicted by the heckler model @xcite . the limits on @xmath7 derived at @xmath395 using the @xmath10 , @xmath11 , @xmath103 , @xmath26 , @xmath27 , @xmath15 and @xmath16 background spectra @xcite will not be significantly affected . we also note that the high energy limits on the detectability of individual holes , which are derived at @xmath95 , are not sensitive to behaviour around @xmath322 but the limits on @xmath7 , which are derived at @xmath396 , may be . this is because at @xmath396 most particles in the integrated spectra from a pbh background are qcd jet products from @xmath397 black holes emitting in the present era @xcite . the galactic antiproton limit is less influenced by the behaviour around @xmath322 because most of the postgalactic pbh antiprotons are generated in somewhat higher energy qcd jets , due to the greater rest mass of the antiproton @xcite . in future work a more precise numerical modeling of the spectra at @xmath398 mev using our insights of section iv.k should be pursued . belyanin , kocharovsky , and kocharovsky @xcite have proposed an unrelated photosphere model , in which charged particles from the black hole predominantly pions from the decay of hawking radiated quarks and gluons form a well - defined ` collisionless ' plasma at @xmath160 . the hydrodynamical behavior of this plasma is assumed to be maintained by collective wave - particle interactions and/or a self - induced turbulent magnetic field , both of which the authors postulate can arise from anisotropies in the charged particle distributions . particle - particle collisions themselves are insufficient to maintain this hydrodynamical regime . belyanin , kocharovsky , and kocharovsky claim that there is sufficient time for plasma wave turbulence to develop as the particles stream away from a high temperature black hole . explicitly , they find that , when @xmath399 , the size of the plasma ( taken to be the distance from the black hole ) is greater than the debye radius @xmath400 ( effectively the inverse of the langmuir oscillation frequency or plasma frequency generated by the charged particle separation ) , and that @xmath400 is greater than the average separation between charged particles . the authors apply magneto - hydrodynamical equations for a plasma magnetic field in the case where the magnetic field is maximal , i.e. where the energy is equipartitioned between particle kinetic energy and the magnetic field . they argue that , for @xmath401 , synchrotron radiation and electromagnetic cascading in the self - induced magnetic field will produce an observable mev @xmath4-ray burst lasting @xmath402 . they envision three possible sources for the induced turbulent magnetic field . these involve the development , as the charged particles move away from the black hole , of unspecified plasma - wave turbulence or strong internal shocks . suggested shock generators are @xmath309 pairs created by electromagnetic cascade in the postulated magnetic field of the emitted particles , or collisions of emitted particles with the interstellar medium . the belyanin , kocharovsky , and kocharovsky scenario makes the following assumptions : that the charged particles interact sufficiently to form a plasma ; that a substantial magnetic field is generated by anisotropies in the charged particle distributions ; and that significant turbulence arises in this field . in analyzing this scenario , we find that a number of the model components have been treated incorrectly . we conclude that a self - induced photosphere of this type should not occur around the black hole . first , langmuir oscillations are longitudinal oscillations that arise in a plasma when the positively charged particles move collectively in the opposite direction to or independently of the negatively charged particles . the mutual attraction between the separated positive and negative charges provides a restoring force that constrains the charged particles to oscillate around their equilibrium positions in the rest frame of the plasma . the oscillations can propagate along this axis of particle motion . for example , langmuir oscillations can occur in an ionic plasma placed in an external magnetic field . however , in the case of particles streaming away from a black hole , we note that there is no initial asymmetry in the distribution or response of positive and negative charges for every particle , a neutral black hole stochastically emits on a short timescale an antiparticle of equal mass and opposite electric charge @xcite . any initial black hole charge is lost on a timescale much less than the age of the universe for @xmath403 @xcite . the remaining random charge fluctuations are of order the planck charge @xmath404 and therefore negligible @xcite . additionally , a neutral black hole has no intrinsic magnetic field . if emitted particles decay , total charge symmetry is preserved . in particular , the net electric charge of the final products of a qcd jet initiated by a decaying quark is equal to the negligibly small charge of the original quark , i.e. @xmath405 or @xmath406 . moreover , self - induced qed turbulence is not observed in accelerator jets . hence the outflow from a black hole remains essentially electrically neutral on all scales and langmuir oscillations should not arise from charge asymmetries in the flux distribution . although we do not expect a self - induced magnetic field to develop as the emitted particles stream away from the black hole , let us for the moment assume that collective wave - particle interactions and/or a turbulent magnetic field do arise in the charged particle flux when the hole is placed in an external magnetic field of sufficient strength and/or turbulence . pions form via hadronization in qcd jets at a distance from the black hole of roughly @xmath407 taking @xmath408 for @xmath326 and @xmath235 quarks , with faster pions forming farther from the black hole and more coaxially than slower ones . the charged pions will decay at a distance @xmath409 where @xmath410 is the energy of the charged pion and @xmath411 is the charged pion lifetime . the average energy of the pions produced by the black hole is @xcite @xmath412 this value is distorted by the high energy @xmath413 particles but most pions in the instantaneous number flux are ` slow ' with energies @xmath414 @xcite . if langmuir oscillations do arise in the charged pion flux , their frequency in the rest frame of the pion plasma is @xmath415 where the prime indicates quantities in the plasma rest frame , @xmath416 is the electron charge in electrostatic units and @xmath417 is the charged pion number density . transforming into the rest frame of the black hole , the langmuir frequency is @xmath418 ( see , for example , ref . this disagrees with eq . ( 10 ) of belyanin , kocharovsky , and kocharovsky which omits a factor of @xmath419 . in the black hole frame , the pion number density and flux are @xmath420 and @xmath421 , respectively @xcite . in the case of the slow pions with @xmath422 , the first constraint for the hydrodynamical regime that the debye radius @xmath423 ( which corresponds to @xmath424 , the distance at which the kinetic energy of a plasma particle balances its electrostatic potential energy in the plasma rest frame ) be less than the characteristic size of the plasma @xmath108 is satisfied when @xmath425 gev . for the fast pions with @xmath426 , however , @xmath427 is an order of magnitude greater than @xmath108 at all @xmath104 and this constraint is never satisfied . in fact , at any given @xmath104 , the @xmath428 constraint is satisfied only by pions with energy @xmath429 gev . thus the claim by belyanin , kocharovsky , and kocharovsky that @xmath428 is satisfied by all pions when @xmath104 is greater than a few gev is not supported by our analysis . the second constraint for the hydrodynamical regime that the debye radius be greater than the average separation between charged pions in the plasma rest frame , i.e. @xmath430 corresponds to @xmath431 in the black hole rest frame . the @xmath432 factor is omitted by belyanin , kocharovsky , and kocharovsky . in the case of the slow pions , this constraint is satisfied when the pions reach a distance from the black hole of @xmath433 in the case of the fast pions , it is satisfied when the pions reach a distance @xmath434 recall , however , that the fast pions never satisfy the @xmath435 constraint at any @xmath108 . for all pions , we also have the third requirement that the first and second constraints can be satisfied only if @xmath108 is greater than the distance at which the pions form . thus the region in which all three constraints are satisfied is @xmath436 for slow pions and non - existent for fast pions . repeating the above analysis for electrons and positrons , representative energies are @xmath437 for the fast @xmath65 and @xmath438 for the slow @xmath65 @xcite . the total @xmath65 flux produced by a black hole is @xmath439 @xcite . for the slow @xmath65 , it follows that @xmath440 when @xmath441 . for the fast @xmath65 , @xmath442 is never satisfied at any @xmath104 and the @xmath440 constraint is satisfied only by those @xmath65 with energy @xmath443 . the second constraint , @xmath444 , is satisfied when the slow @xmath65 have reached a distance @xmath445 from the black hole and when the fast @xmath65 have reached a distance @xmath446 . however , the @xmath65 produced by the @xmath447 decay chain form at a distance @xmath448 , which is much greater than @xmath449 . combining these results , the region in which all three constraints are satisfied is @xmath450 for slow @xmath65 . for fast @xmath65 , there is no region in which all three constraints are satisfied . in the above discussion , we assumed that the strength of the ambient magnetic field was sufficient to generate hydrodynamical behavior . let us now estimate the strength of the external magnetic field required to cause significant longitudinal separation of positive and negatives charges as they move radially away from the black hole . as we saw in eq . , the predominantly radially - moving particles may have a transverse velocity component @xmath451 in the black hole frame . using the force equations @xmath452 and @xmath453 , the separation along the radial direction induced by the magnetic field is given by @xmath454 , implying @xmath455 @xmath456 certainly the galactic magnetic field , which is of order @xmath457 , is too weak to produce significant charge separation and hence can not generate langmuir oscillations . additionally , the magnetic field strength required for significant separation would appear to be unrealistically large in any astrophysical context . if we consider the pions , electrons and positrons to be generated in qcd jets of opening angle @xmath458 , the transverse velocity may be as large as @xmath459 and so the maximum separation is given by @xmath460 for @xmath461 and @xmath462 smaller for @xmath65 . at accelerator energies , @xmath463 @xcite . the maximum magnetic field - induced separation is then greater than the average separation between particles , @xmath464 , at @xmath465 @xmath466 in this case , turbulence may be generated at high temperatures in a strong external magnetic field . however , only a small fraction of the emitted particles would have sufficient transverse velocity and the relevant orientation to satisfy the conditions to participate in langmuir oscillations . we have already shown above that , once started , only the low energy @xmath461 and @xmath65 can possibly sustain an mhd regime . we conclude that a _ self - induced _ magnetic field and mhd regime should not arise in the charged emitted particles as they stream away from the black hole . hence no photosphere will be generated by this mechanism . if the black hole is embedded in an _ external _ magnetic field of sufficient strength and/or turbulence , we find that the mhd regime extends over a smaller region than belyanin , kocharovsky , and kocharovsky envisage and can involve only the lowest energy emitted particles , weakening the effect . rees has also proposed a model in which the high energy electrons emitted by an expiring black hole form an expanding conducting shell which will be braked by any ambient galactic magnetic field @xcite . however , macgibbon has found that the rees mechanism , when re - analyzed including qcd decays , will produce a gamma - ray burst with a detection probability at earth of much less than @xmath136 @xcite . d. cline , matthey and otwinowski have proposed that galactic pbhs may explain the shortest - lived gamma - ray bursts @xcite . they suggest two mechanisms for burst generation : plasma formation at @xmath467 , together with the mhd instability proposed by belyanin , kocharovsky , and kocharovsky @xcite ; or ` fireball ' formation at the qcd threshold @xmath322 , as proposed by cline and hong @xcite . as we have seen , the belyanin , kocharovsky , and kocharovsky mechanism is not viable in the absence of a large external magnetic field . the second mechanism postulates that the emitted particles undergo a first - order phase transition around @xmath322 , accompanied by rapid loss of the remaining black hole mass via a hagedorn - type exponential increase in the number of hadronic degrees of freedom at @xmath322 , without the black hole temperature climbing above @xmath1 . these hadrons are assumed to be directly emitted by the black hole , and not the result of decays . the hagedorn model was suggested by the exponential increase in hadronic resonances observed in early accelerator experiments at these energies . however , the resonances have long since been understood in terms of the more fundamental quark model , and accelerator energies well above @xmath1 have long since been achieved . the standard hawking picture , in which the black hole emits only those particles which are fundamental on the scale of the black hole and which was discussed in detail in section iv.k , is consistent with current accelerator observations . thus a @xmath397 hole should directly emit quarks and gluons , which then hadronize at a distance @xmath468 from the black hole , rather than an explosive hagedorn - type particle distribution . similarly an @xmath469 black hole should emit pions as fundamental particles . furthermore , in the hagedorn picture the exponential distribution of resonances are possible particle states but whether the states are occupied is determined by the available energy . as we have seen in section iv.k , the energy per emitted quark or gluon around @xmath322 is only @xmath337 . dense occupation of the hagedorn resonances would require a direct coupling of the black hole mass to the hagedorn states . if the black hole is rapidly losing mass into the hagedorn spectrum , however , eq . dictates that the temperature must rapidly climb above @xmath1 . this would violate the condition of the original hagedorn model @xcite in which the exponential growth in the density of states results from the temperature being held constant . hence , based on the observed behavior in accelerators , there should be no rapid increase in the black hole evaporation rate as @xmath470 and no gamma - ray photosphere burst of the type envisioned by cline and hong . the ` fireball ' model proposed by moss @xcite , in which a phase - transition at the @xmath1 threshold generates a confining bubble around the black hole , is also not consistent with accelerator observations . note that the cline and hong ` fireball ' model is not related to the heckler bremsstrahlung - induced photosphere . a heckler - type photosphere involves only interactions of particles after emission and is not accompanied by a rapid increase in the hawking evaporation rate . although our analysis rules out photospheres in such models , we can not exclude the possibility that black hole explosions could generate short - period gamma - ray bursts by some other as yet unconsidered mechanism or by placement , for example , in unusual magnetic field configurations . the possibility that one might detect black hole explosions ( and thereby test the hawking theory ) at the present epoch is so important that all such scenarios are worth exploring . in particular , it is important to examine whether short - period gamma - ray bursts have other characteristics expected of a pbh population , independent of knowledge of the specific burst production mechanism . after careful examination of the heckler photosphere model , we find that the emitted particles are not expected to interact sufficiently to form a qed or qcd photosphere around an evaporating black hole . key to determining the fate of the emission is understanding the causality constraint and scatterings which occur within a distance @xmath118 or @xmath471 of the hole . our main point is that no qed photosphere can form around a black hole because a particle must be causally connected to any particle it interacts with and requires a distance @xmath0 to form in each bremsstrahlung interaction . this means that each particle is highly unlikely to interact and generally undergoes at most one net interaction as it streams away from the hole . neither heckler nor cline , mostoslavsky , and servant include these effects . instead they envision a scenario in which each emitted particle undergoes a sequence of interactions . in heckler s model these start at a radius less than @xmath239 and continue to a radius greater than @xmath239 ; in the cline , mostoslavsky , and servant simulations , they start at a radius much less than @xmath239 and continue to @xmath239 . furthermore , the modification required to their models is not a simple one , since the causality and formation distance constraints and a corrected geometrical description must go into the analysis at the start and can not simply be added later . additional qcd phenomenological arguments , based on the limited energy per initial particle and availability of qcd final states , also rule out qcd photosphere ( chromosphere ) development around @xmath322 . although we find that photosphere formation is not supported , an emitted particle may occasionally undergo one net qed interaction within a distance @xmath118 of the black hole or one net qcd interaction within @xmath471 . however , this interaction could only very slightly modify the observable @xmath10 , @xmath11 , @xmath103 , @xmath26 , @xmath27 , @xmath15 and @xmath16 spectra from a pbh background or an individual high - temperature black hole . the spectra remain essentially those of the standard hawking model , with little change to the detection probability . we also find that interactions are insufficient to support other models for photosphere formation or enhanced pbh signals proposed in the literature , unless the black hole is located in an unusual ambient environment or standard model physics breaks down . in this paper , we have not addressed the emission expected from higher - dimensional black holes which may be created at tev or higher energies in accelerators @xcite or cosmic ray interactions @xcite . our analysis implies that the spectra which were calculated , in for example refs . @xcite , assuming heckler - type interactions between the particles emitted in the final burst of evaporation from a higher - dimensional black hole will need to be modified . in particular , the issues raised in sections iv.b , d , f , g , h and k will need to be applied to the analysis of the spectra from higher - dimensional black holes . in the case of accelerator experiments , the relevant condition to investigate is whether @xmath472 is approached , because even one interaction per particle may modify the spectra significantly compared with experimental precision . a number of the issues we raise were independently recently included in the numerical simulation of the decay of cern lhc tev - scale black holes by alig , drees and oda @xcite . they concluded that scattering between partons emitted by the higher - dimensional black hole is insufficient to form a thermalized chromosphere . although their methodology was restricted to scattering between partons which approach each other and included a lower cutoff of @xmath473 in the exchanged transverse momentum , their results provide insight into the applicability of the heckler scenario to tev - scale higher - dimensional black holes . our discussion of geometrical and causality considerations in section iv.f and in our accompanying paper @xcite may be viewed as a partial justification of the approximation used in @xcite of only including scattering between partons which approach each other . the authors thank andrzej czarnecki , manuel drees , joseph kapusta , graham thompson and bryan webber for useful discussions . jhm is grateful to queen mary , university of london , the fermilab astrophysics group and the university of cambridge for hospitality received at various points during this work . bjc is grateful to the research center for the early universe at the university of tokyo for hospitality received during its completion . dnp thanks the natural sciences and engineering research council of canada for financial support . e. bugaev , p. klimai , and v. petkov , `` photon spectra from final stages of a primordial black hole evaporation in different theoretical models '' to appear in proceedings of 30th international cosmic ray conference , merida , mexico , july 2007 arxiv.0706.3778 . l. d. landau and i. j. pomeranchuk , dokl . sssr * 92 * , 535 ( 1953 ) ; l. d. landau and i. j. pomeranchuk , dokl . sssr * 92 * , 735 ( 1953 ) ; a. b. migdal , phys.rev . * 103 * , 1811 ( 1956 ) ; a. b. migdal , sov . jetp * 5 * , 527 ( 1957 ) . h. satz , `` thermal hadron production by qcd hawking radiation '' to appear in proceedings of the third international workshop on the critical point and onset of deconfinement , florence , italy , 2006 hep - ph/0612151 . n. brambilla and a. vairo , `` quark confinement and the hadron spectrum '' lectures given at the 13th annual hampton university graduate studies summer school at the continuous electron beam facility , newport news , virgina , 1998 hep - ph/9904330 .	several authors , most notably heckler , have claimed that the observable hawking emission from a microscopic black hole is significantly modified by the formation of a photosphere around the black hole due to qed or qcd interactions between the emitted particles . in this paper we analyze these claims and identify a number of physical and geometrical effects which invalidate these scenarios . we point out two key problems . first , the interacting particles must be causally connected to interact , and this condition is satisfied by only a small fraction of the emitted particles close to the black hole . second , a scattered particle requires a distance @xmath0 for completing each bremsstrahlung interaction , with the consequence that it is improbable for there to be more than one complete bremsstrahlung interaction per particle near the black hole . these two effects have not been included in previous analyses . we conclude that the emitted particles do not interact sufficiently to form a qed photosphere . similar arguments apply in the qcd case and prevent a qcd photosphere ( chromosphere ) developing when the black hole temperature is much greater than @xmath1 , the threshold for qcd particle emission . additional qcd phenomenological arguments rule out the development of a chromosphere around black hole temperatures of order @xmath1 . in all cases , the observational signatures of a cosmic or galactic halo background of primordial black holes or an individual black hole remain essentially those of the standard hawking model , with little change to the detection probability . we also consider the possibility , as proposed by belyanin et al . and d. cline et al . , that plasma interactions between the emitted particles form a photosphere , and we conclude that this scenario too is not supported . phys . rev . * d 78 * , 064043 ( 2008 )
in this paper , we explore the dynamics of monodromies of fibrations on a hyperbolic , fibered @xmath1-manifold , called the _ magic @xmath1-manifold _ @xmath0 , which is the exterior of the @xmath1 chain link @xmath9 , see figure [ fig_3braid](1 ) . we first set some notations , then describe the motivation of our study . let @xmath10 be an orientable surface ( possibly with punctures ) . a homeomorphism @xmath11 is _ pseudo - anosov _ if there exist a pair of transverse measured foliations @xmath12 and @xmath13 and a constant @xmath14 such that @xmath15 then @xmath16 and @xmath17 are called the _ unstable _ and _ stable foliations _ ( or _ invariant foliations _ ) , and @xmath18 is called the _ dilatation _ of @xmath19 . let @xmath20 be the mapping class group of @xmath10 , that is @xmath20 is the group of isotopy classes of orientation preserving homeomorphisms on @xmath10 fixing punctures setwise . a mapping class @xmath21 is called _ pseudo - anosov _ if @xmath22 contains a pseudo - anosov homeomorphism @xmath11 as a representative . the topological entropy @xmath23 of @xmath19 is equal to @xmath24 , and @xmath23 attains the minimal entropy among all homeomorphisms which are isotopic to @xmath19 , see ( * ? ? ? * expos 10 ) . in the case @xmath25 $ ] , we denote by @xmath26 and @xmath27 , the dilatation @xmath28 and topological entropy @xmath29 . we take an element @xmath21 . let @xmath30 be its mapping torus , i.e , if @xmath19 is a representative of @xmath22 , then @xmath31 where @xmath32 identifies @xmath33 with @xmath34 for @xmath35 and @xmath36 . such a @xmath19 is called the _ monodromy _ of @xmath30 . the vector field @xmath37 on @xmath38 induces a flow @xmath39 on @xmath30 , which is called the _ suspension flow_. the hyperbolization theorem by thurston @xcite tells us that a @xmath1-manifold @xmath40 which is homeomorphic to @xmath30 admits a hyperbolic structure if and only if @xmath22 is pseudo - anosov . the magic manifold @xmath0 is in fact a hyperbolic , fibered @xmath1-manifold , since @xmath0 is homeomorphic to a @xmath41-puncture sphere bundle over the circle with the pseudo - anosov monodromy as in figure [ fig_3braid](2 ) ( see lemma [ lem_summary](1 ) ) . chain link @xmath9 . ( 2 ) braided link @xmath42 . our convention of the orientation is that the meridian of a component of the link is chosen as in ( 3).,width=288 ] in a paper @xcite thurston introduced a norm @xmath43 on @xmath44 for hyperbolic @xmath1-manifolds and proved that the unit ball @xmath45 with respect to the thurston norm @xmath43 is a compact , convex polyhedron . when @xmath40 is homeomorphic to a hyperbolic , fibered @xmath1-manifold @xmath30 , he gave a connection between the thurston norm @xmath43 and fibrations on @xmath40 . in particular if such a @xmath1-manifold @xmath40 has the second betti number @xmath46 which is greater than @xmath47 , he proved that there exists a top dimensional face @xmath48 on @xmath49 , called a _ fibered face _ such that every integral class @xmath4 of @xmath50 which is in the open cone over @xmath48 corresponds to a fiber @xmath2 of the fibration associated to @xmath4 with a monodromy @xmath51 , that is @xmath52 is homeomorphic to @xmath40 , where @xmath53 $ ] . such an integral class @xmath4 is called a _ fibered class_. since @xmath53 $ ] is pseudo - anosov for each fibered class @xmath4 , @xmath40 provides infinitely many pseudo - anosovs on fibers with distinct topological types . a theorem by fried @xcite asserts that the monodromy @xmath51 and the ( un)stable foliation @xmath54 of @xmath55 can be described by using the suspension flow @xmath39 and the suspension of the ( un)stable foliation of @xmath19 . a question we would like to pose is that how about practical constructions of @xmath56 and @xmath54 for each fibered class @xmath4 . the theorem by fried does not give us concrete descriptions of them . e. hironaka gave concrete descriptions of the monodorimies of fibrations associated to sequences of fibered classes on some class of hyperbolic fibered @xmath1-manifolds , see @xcite . however no one constructed explicitly the monodromy of the fibration associated to each fibered class on a single hyperbolic , fibered @xmath1-manifold @xmath40 with @xmath57 . in this paper we describe them concretely for the magic manifold @xmath0 . the motivation of our study comes from minimal dilatations on pseudo - anosovs and their asymptotic behaviors . we fix a surface @xmath10 , and consider the set of dilatations of pseudo - anosovs on @xmath10 , @xmath58 arnourx - yoccoz and ivanov observed that for any constant @xmath59 , there exist finite elements @xmath60 so that @xmath61 , see @xcite . in particular , there exists a minimum @xmath62 of @xmath63 . let @xmath64 be a closed surface of genus @xmath7 , and @xmath65 a closed surface of genus @xmath7 removing @xmath66 punctures . we let @xmath67 and @xmath68 . we denote by @xmath69 , an @xmath70-punctured disk . a mapping class @xmath71 defines a mapping class @xmath72 fixing a puncture and vice versa . moreover @xmath22 is pseudo - anosov if and only if @xmath73 is pseudo - anosov . in this case the equality @xmath74 holds . thus @xmath75 is equal to the minimal dilatations among pseudo - anosov elements @xmath72 fixing a puncture . in particular we have @xmath76 . the minimal dilatation problem is to determine an explicit value of @xmath62 , and to identify a pseudo - anosov element @xmath77 which achieves @xmath62 ( i.e , minimizer of @xmath62 ) . naive but natural question is this : what does the pseudo - anosov homeomorphism @xmath19 on @xmath10 which achieves @xmath62 look like ? in other words , what does a train track representative of @xmath78 $ ] ( which enables us to describe the dynamics of @xmath79 ) look like ? @xmath80 some of the minimal dilatations are already determined . also there are partial results . for example , @xmath81 is computed in @xcite , but an explicit value of @xmath5 is not known for @xmath82 . if we denote by @xmath6 , the minimal dilatation of pseudo - anosovs defined on @xmath64 with orientable invariant foliations , then explicit values of @xmath6 for @xmath83 except @xmath84 are known , see @xcite and @xcite . the minimal dilatation on an @xmath70-punctured disk , @xmath75 is determined for @xmath85 , see @xcite . the asymptotic behaviors of the minimal dilatations are shown in the left column of table [ table_section ] . here @xmath86 means that there exists a constant @xmath87 which does not depend on @xmath7 so that @xmath88 . as we can see from table , we have @xmath89 and @xmath90 , but a result by tsai @xcite says that the situation in the case @xmath91 is quite different from the case @xmath92 or @xmath47 . in the right column of table [ table_section ] , the smallest known upper bounds of the minimal dilatation . we give a precise condition @xmath93 in ( u6 ) in the following . [ thm_boundary ] suppose that @xmath91 satisfies _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ @xmath93 @xmath94 or @xmath95 for each @xmath96 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ then @xmath97 in particular , if @xmath98 is prime , then @xmath7 enjoys @xmath93 , and hence ( [ equation_tsai ] ) holds . the upper bounds ( u1)(u6 ) are proved by examples of sequences of pseudo - anosovs . the magic manifold @xmath0 is involved in these upper bounds . there exists a sequence of fibered classes of @xmath0 corresponding to each of ( u1 ) , @xmath99 , ( u6 ) . the projection onto a fibered face of the sequence corresponding to ( u6 ) converges to a single point which lies on the boundary of a fibered face of @xmath0 . on the other hand , the projection of other sequence converges to some point in the interior of the fibered face of @xmath0 . an interesting feature is the following . the mapping torus @xmath30 of each example @xmath22 which appears in the sequences is either @xmath0 or the fibration of @xmath30 comes from a fibration of @xmath0 by dehn filling cusps along the boundary slopes of a fiber . this is also true for known minimizers of the minimal dilatations @xmath81 , @xmath75 for @xmath100 and @xmath6 for @xmath83 except @xmath84 . these results say that the topological types of fibers of fibrations on @xmath0 are surprisingly full of variety . however , no explicit constructions of sequences of pseudo - anosovs needed for the proofs of ( u1)(u6 ) except ( u4 ) were given so far . also an explicit example of a minimizer of @xmath8 was not given . in this paper we prove the following which allows us to construct pseudo - anosovs in question explicitly . [ thm_algorithm ] we have algorithms to construct the followings . for each fibered class @xmath4 of @xmath0 , 1 . the monodromy @xmath51 of the fibration on @xmath0 associated to @xmath4 , and 2 . in the case @xmath4 is primitive , a train track representative @xmath101 of @xmath53 $ ] whose incidence matrix is perron - frobenius . in @xcite oertel constructs branched surfaces which carry fibers of fibrations on hyperbolic , fibered @xmath1-manifolds . in the proof of theroem [ thm_algorithm](1 ) , we construct branched surfaces @xmath102 and @xmath103 following @xcite which carry fibers of fibrations associated to fibered classes on @xmath0 . it is well - known that a train track representative @xmath101 as in theorem [ thm_algorithm](2 ) can recover a pseudo - anosov homeomorphism which represents @xmath104 , and it serves the monodromy @xmath56 of the fibration associated to @xmath4 . however we do not need the claim ( 2 ) for the proof of ( 1 ) . we can construct the both fiber @xmath2 and monodromy @xmath51 in an explicit and combinatorial way . let @xmath105 be the manifold obtained from @xmath0 by dehn filling one cusp along the slope @xmath106 . see figure [ fig_3braid](3 ) for our convention of the orientation . as a consequence of theorem [ thm_algorithm ] , we can give constructive descriptions of monodromies of fibrations associated to any fibered class on the hyperbolic , fibered manifolds @xmath105 for infinitely many @xmath106 . for example , we can do them for whitehead sister link exterior @xmath107 , the the simplest @xmath1-braided link exterior @xmath108 and the whitehead link exterior @xmath109 . in particular , we can construct the following pseudo - anosov homeomorphisms explicitly : * a minimizer of @xmath8 , see example [ ex_ori79 ] . * conjectural minimizers of @xmath6 for @xmath110 , see ( * ? ? ? * question 6.1 ) . see also remark [ rem_hironaka - san ] and example [ ex_lt ] . * conjectural minimizers of @xmath6 for large @xmath7 such that @xmath111 , see ( * ? ? ? * conjecture 1.12(2 ) ) . see also examples [ ex_lt ] , [ ex_ori79 ] , [ ex_ori15 ] . * conjectural minimizer of @xmath5 for large @xmath7 , see ( * ? ? ? * conjecture 1.12(1 ) ) . see also example [ ex_ori79 ] . * conjectural minimizers of @xmath112 for large @xmath70 , see ( * ? ? ? * conjecture 1.13 ) . see also example [ ex_whitehead ] . * sequences of pseudo - anosov homeomorphisms to give the smallest known upper bounds ( u1)(u6 ) in table [ table_section ] . [ rem_hironaka - san ] hironaka gave the first explicit construction of the orientable train track representative on @xmath64 for @xmath110 whose dilatation equals the conjectural minimum @xmath6 for such a @xmath7 , see @xcite . hironaka also constructed explicitly the infinite subsequence of pseudo - anosov homeomorphisms defined on @xmath64 with some condition on @xmath7 to prove the upper bound ( u1 ) and ( u2 ) , see @xcite . find the word which represents the mapping class @xmath53 $ ] for each fibered class @xmath4 of @xmath0 by using the standard generating set on @xmath113 . its word length could be long , but would be represented by a simple word , see remark [ rem_delta0 ] . develop the methods given in section [ section_construction ] and prove the same claim in theorem [ thm_algorithm ] for some class of hyperbolic , fibered @xmath1-manifolds . the paper is organized as follows . in section [ section_preliminarlies ] , we review basic facts on train tracks , thurston norm and clique polynomials . we also review some properties of the magic manifold . in section [ section_construction ] , we prove theorem [ thm_algorithm ] . in its proof , we construct the directed graph @xmath114 with a metric on the set of edges , which is induced from the train track representative @xmath101 of @xmath104 . such a directed graph @xmath114 captures the dynamics of both @xmath51 and @xmath101 . then we construct the curve complex @xmath115 induced from @xmath114 , which is an undirected , weighted graph on the set of vertices . such curve complexes are recently studied by mcmullen @xcite . in our setting , @xmath115 gives us some insight into what the train track representative @xmath101 looks like . in section [ section_catalogue ] , we exhibit some subsequences of pseudo - anosovs which can be used in the proof of the upper bounds ( u1)(u6 ) . we find that the types of curve complexes in each subsequence are fixed . these curve complexes give us some hints to know what the pseudo - anosovs with the smallest dilations look like . * acknowledgment . * i would like to thank hideki miyachi , mitsuhiko takasawa and hiroyuki minakawa . h. miyachi and m. takasawa gave me valuable comments on this paper . minakawa gave a series of lectures on his work at osaka university in 2004 . i learned many things on pseudo - anosovs and pseudo - anosov flows during his course . theorem [ thm_algorithm](1 ) is inspired by his construction of pseudo - anosovs @xcite . definitions and basic results on train tracks are contained in @xcite . see also @xcite . in this section , we recall them for convenience . throughout the paper , surfaces are orientable . let @xmath116 be a surface with possibly punctures or boundary . let @xmath117 be a branched @xmath47-submanifold on @xmath116 . we say that @xmath117 is a _ train track _ if 1 . @xmath117 is a smooth graph such that the edges are tangent at the vertices , i.e , @xmath117 looks as in figure [ fig_local](1 ) near each vertex of @xmath117 , 2 . each component of @xmath118 is a disk with more than @xmath1 cusps on its boundary or an annulus with more than @xmath47 cusp on one boundary component and with no cusps on the other boundary component ( i.e , the other boundary component is the one of @xmath119 or a puncture of @xmath116 . ) see figure [ fig_seedabcd](1 ) for an example of a train track on @xmath120 . two edges of @xmath117 which are tangent at some vertex make a _ cusp _ , see figure [ fig_folding](left ) . associated to the train track @xmath117 , we can define a fibered neighborhood @xmath121 whose fibers are segments given by a retraction @xmath122 . the fibers in this case are called _ ties _ , see figure [ fig_local](2 ) . let @xmath123 be a measured foliation on @xmath116 . we say that @xmath123 _ is carried by _ @xmath117 if @xmath123 can be represented by a partial measured foliation whose support is @xmath124 and which is transverse to the ties . let @xmath125 be a train track on @xmath116 . we say that @xmath125 _ is carried by _ @xmath117 if @xmath125 is isotopic to a train track @xmath126 which is contained in @xmath124 and which is transverse to the ties ( said differently , every smooth edge path on @xmath126 is transverse to the ties ) . let @xmath127 be a homeomorphism . a train track @xmath117 is _ invariant under _ @xmath128 $ ] if @xmath129 is carried by @xmath117 , that is , @xmath129 is isotopic to some train track @xmath126 which satisfies above . in this case , folding edges of @xmath130 near cusps repeatedly ( see figure [ fig_folding ] for a folding map ) , in other words , collapsing @xmath130 onto @xmath117 smoothly yields a map @xmath131 such that @xmath132 maps vertices to vertices , and @xmath132 is locally injective at any points which do not map into vertices . such a @xmath131 is called a _ train track representative _ of @xmath128 $ ] . an edge @xmath133 of @xmath117 is said to be _ infinitesimal _ ( for @xmath132 ) if @xmath133 is eventually periodic under @xmath132 , that is @xmath134 for some positive integers @xmath135 and @xmath70 . other edges of @xmath117 is said to be _ real_. let @xmath136 be the number of the real edges of @xmath117 . then we have a @xmath137 non - negative integer matrix @xmath138 , called the _ incidence matrix _ for @xmath131 _ ( with respect to real edges ) _ , where @xmath139 is the number of times so that the image @xmath140 of the @xmath141th real edge passes through the @xmath142th real edge @xmath143 in either direction . also , @xmath131 determines a finite , directed graph @xmath144 by taking a vertex for each real edge of @xmath117 , and then adding @xmath139 directed edges from the @xmath141th real edge @xmath145 to the @xmath142th real edge @xmath143 . in other words , we have @xmath139 directed edges from @xmath145 to @xmath143 if @xmath140 passes through @xmath143 in either direction @xmath139 times . we say that @xmath144 is the _ induced directed graph _ of @xmath131 . a non - negative integer matrix @xmath40 is said to be _ perron - frobenius _ if there exists an integer @xmath146 such @xmath147 is positive , that is each entry of @xmath147 is positive . in this case , the spectral radius of @xmath40 is given by the largest eigenvalue of @xmath40 called the _ perron - frobenius eigenvalue _ , see @xcite . the following theorem is well known . [ thm_pape ] if @xmath148 is a pseudo - anosov homeomorphism , then there exists a train track @xmath117 on @xmath116 which carries the unstable foliation @xmath16 of @xmath19 and a train track representative @xmath131 of @xmath25 $ ] . such a representative @xmath131 satisfies that the incidence matrix @xmath149 is perron - frobenius and its perron - frobenius eigenvalue is exactly equal to @xmath28 . conversely , if @xmath127 is a homeomorphism and if @xmath131 is a train track representative of @xmath150 $ ] such that its incidence matrix @xmath149 is perron - frobenius , then @xmath22 is pseudo - anosov whose dilatation @xmath26 equals the perron - frobenius eigenvalue of @xmath149 , see bestvina - handel ( * ? ? ? * section 3.4 ) we review the basic results on thurston norm and the relation between thurston norm and hyperbolic , fibered @xmath1-manifolds developed by thurston , fried , matsumoto and mcmullen . let @xmath40 be an oriented hyperbolic @xmath1-manifold possibly @xmath151 . we recall thurston norm @xmath152 . for more detail , see @xcite . let @xmath153 be a finite union of oriented , connected surfaces . we define @xmath154 to be @xmath155 thurston norm @xmath43 is defined for an integral class @xmath156 by @xmath157 where the minimum is taken over all oriented surfaces @xmath116 embedded in @xmath40 satisfying @xmath158 $ ] . the surface @xmath116 which realizes the minimum is called a _ minimal representative _ of @xmath4 . then @xmath43 admits a unique continuous extension @xmath159 which is linear on rays through the origin . it is known that the unit ball @xmath160 with respect to thurston norm @xmath43 is a compact , convex polyhedron @xcite . let @xmath48 be any top dimensional face on the boundary @xmath49 of thurston norm ball . we denote by @xmath161 , the cone over @xmath48 with the origin , and we denote by @xmath162 , the interiors of @xmath161 . thurston proved in @xcite that if @xmath40 is a surface bundle over the circle and if @xmath116 be any fiber of the fibration on @xmath40 , then there exists a top dimensional face @xmath48 on @xmath49 so that @xmath163 $ ] is an integral class of @xmath164 . moreover for any integral class @xmath165 , its minimal representative @xmath2 becomes a fiber of a fibration on @xmath40 , and @xmath2 is unique up to isotopy along flow lines . such a face @xmath48 is called an _ fibered face _ and an integral class @xmath165 is called a _ fibered class_. thus , if the second betti number @xmath46 is greater than @xmath47 , then the single @xmath1-manifold @xmath40 provides infinitely many pseudo - anosovs on surfaces with different topological types . let @xmath48 be a fibered face of @xmath40 . if @xmath165 is primitive and integral , then the minimal representative @xmath2 is a connected fiber of the fibration associated @xmath4 . the mapping class @xmath53 $ ] of the monodromy @xmath51 of its fibration is pseudo - anosov ( since @xmath40 is hyperbolic ) . we define the dilatation @xmath166 and entropy @xmath167 to be the dilatation and entropy of the pseudo - anosov @xmath104 . the entropy function defined on primitive fibered classes is naturally extended to rational classes . in fact for a rational number @xmath168 and a primitive fibered class @xmath4 , the entropy of @xmath169 is defined to be @xmath170 . [ thm_fried1 ] the function given by @xmath171 for each rational class @xmath165 extends to a real analytic convex function on @xmath162 . the restriction @xmath172 is a strictly convex function which goes to @xmath173 toward the boundary of @xmath48 . by properties of @xmath43 and @xmath174 , we see that the _ normalized entropy function _ @xmath175 is constant on each ray in @xmath162 through the origin . we choose @xmath25 \in \mathrm{mod}(\varsigma)$ ] , and we consider the mapping torus @xmath30 with the suspension flow @xmath39 . hereafter we fix the orientation of @xmath10 so that its normal direction coincides with the flow direction of @xmath39 . for @xmath176 , we define @xmath177 to be the image of @xmath178 under the projection @xmath179 . [ thm_fried2 ] let @xmath148 be a pseudo - anosov homeomorphism with stable and unstable foliations @xmath17 and @xmath16 on an oriented surface @xmath116 , and let @xmath25 $ ] . let @xmath180 and @xmath181 denote the suspension of @xmath17 and @xmath16 by @xmath19 . if @xmath48 is a fibered face on @xmath30 with @xmath163 \in int(c_{\omega})$ ] , then for any minimal representative @xmath2 of any fibered class @xmath165 , we can modify @xmath2 by isotopy which satisfies the followings . 1 . @xmath2 is transverse to the flow @xmath39 , and the first return map @xmath3 is precisely the pseudo - anosov monodromy @xmath51 of the fibration on @xmath30 associated to @xmath4 . moreover @xmath2 is unique up to isotopy along flow lines . 2 . the stable and unstable foliations of @xmath55 are given by @xmath182 and @xmath183 . following @xcite , we introduce _ flowbands _ in @xmath30 . [ def_flowband ] let @xmath184 and @xmath185 be embedded arcs in @xmath30 . suppose that @xmath184 and @xmath185 are transverse to @xmath39 . we say that @xmath184 _ is connected to _ @xmath185 ( _ with respect to _ @xmath39 ) if there exists a positive continuous function @xmath186 such that for any @xmath187 , we have * @xmath188 , @xmath189 for @xmath190 , and * the map @xmath191 given by @xmath192 is a homeomorphism . the _ flowband _ $ ] is defined by @xmath194=\ { \phi^t(x)\ \|\ x \in j_1,\ 0 \le t \le \frak{t}(x)\}.\ ] ] flowbands are used to build branched surfaces in section [ subsection_branched ] . we review some results on clique polynomials , developed by mcmullen @xcite . as we will see in section [ subsection_clique ] , clique polynomials are useful to compute the dilatations of pseudo - anosov monodromies of fibrations on fibered @xmath1-manifolds . let @xmath195 be a finite , directed graph @xmath196 with a metric @xmath197 on the set of edges @xmath198 . the metric @xmath135 specifies the length of each edge . parallel edges and loops are allowed . we sometimes denote @xmath199 by @xmath196 when @xmath135 is obvious . the _ growth rate _ @xmath200 is defined by @xmath201 where @xmath202 is the number of the closed , directed paths in @xmath196 of length @xmath203 . when @xmath204 is an integer for each @xmath205 , we can add @xmath206 new vertices along each edge @xmath133 to obtain a new directed graph @xmath207 with the metric @xmath208 sending each edge to @xmath47 . then we have @xmath209 suppose that @xmath22 is a pseudo - anosov mapping class . let @xmath131 be a train track representative of @xmath22 given in theorem [ thm_pape ] and let @xmath144 be the induced directed graph of @xmath131 . theorem [ thm_pape ] implies that @xmath210 satisfies @xmath211 let @xmath212 be a finite , undirected graph with no loops or parallel edges . let @xmath213 be a weight on the set of vertices @xmath214 . the subset @xmath215 forms a _ clique _ if they span a complete subgraph . ( we allow @xmath216 . ) the _ clique polynomial _ of @xmath217 is defined by @xmath218 where @xmath219 denotes the cardinality of @xmath220 , the weight of @xmath220 is given by @xmath221 , and the sum is over all cliques @xmath220 s of @xmath212 . we sometimes denote the weighted , undirected graph @xmath217 by @xmath212 when @xmath222 is obvious . mcmullen relates the growth rates @xmath200 s to the clique polynomials via the _ curve complexes _ of @xmath199 s . let @xmath223 be a collection of edges which form a closed , directed loop . if @xmath224 never visits the same vertex twice , then @xmath224 is called a _ simple curve_. a _ multicurve _ is a finite union of simple curves such that no two simple curves share a vertex . the _ curve complex _ of @xmath199 is the undirected graph @xmath212 together with the weight @xmath213 , which is obtained by taking a vertex for each simple curve @xmath224 of @xmath196 , and then joining the two vertices @xmath225 and @xmath226 by an edge when @xmath227 is a multicurve of @xmath196 . then the metric @xmath135 on @xmath198 induces the weight @xmath222 on @xmath214 as follows . @xmath228 [ thm_mcmullen ] let @xmath217 be the curve complex of @xmath199 . then @xmath229 is equal to the the smallest positive root of the clique polynomial @xmath230 of @xmath217 . by ( [ equation_dilatation ] ) and theorem [ thm_mcmullen ] , we can compute the dilatations of pseudo - anosovs by using the clique polynomials of curve complexes associated to the pseudo - anosovs . we do not need to compute the characteristic polynomials of the incidence matrices for the dilatations . this observation is due to birman @xcite . in this section , we review some properties on @xmath0 which will be used in the paper . we give orientations of components , @xmath231 , @xmath232 and @xmath233 of the @xmath1 chain link @xmath9 as in figure [ fig_3braid](1 ) . each component bounds oriented @xmath234-punctured disks , @xmath235 , @xmath236 and @xmath237 respectively . we set @xmath238 $ ] , @xmath239 $ ] , @xmath240 \in h_2(n , \partial n ; { \bbb z})$ ] . then @xmath241 becomes a basis of @xmath242 . we denote the class @xmath243 by @xmath244 . thurston norm ball @xmath245 is the parallelepiped with vertices @xmath246 , @xmath247 , @xmath248 and @xmath249 ( ( * ? ? ? * example 3 ) ) , see figure [ fig_face_map](1 ) . ( note that the minimal representative of @xmath250 is taken to be the @xmath1-punctured sphere , embedded in @xmath251 , which contains the point @xmath252 . ) let @xmath42 be the @xmath1 components link in @xmath253 as in figure [ fig_3braid](2 ) , i.e , it is the link obtained from the closed @xmath1-braid @xmath254 together with the braid axis . then @xmath255 is homeomorphic to @xmath256 , see lemma [ lem_summary](1 ) . this implies that @xmath0 is a surface bundle over the circle with a fiber of the @xmath41-punctured sphere . notice that every top dimensional face of @xmath257 is a fibered face , because of the symmetries of @xmath9 . to study monodromies of fibrations on @xmath0 , we can pick a particular fibered face , for example the fibered face @xmath258 with vertices @xmath259 , @xmath260 , @xmath261 and @xmath262 , see figure [ fig_face_map](1 ) . the open face @xmath263 is written by @xmath264 thurston norm @xmath265 of @xmath266 is given by @xmath267 . an integral class @xmath268 is fibered ( i.e , an integral class @xmath244 is in @xmath269 ) if and only if @xmath270 , @xmath271 and @xmath272 are integers such that @xmath273 , @xmath274 , @xmath275 and @xmath276 , see ( [ equation_openface ] ) . any class @xmath277 satisfies @xmath278 . hence we can write such a class @xmath279 by @xmath280 $ ] . then we have @xmath281\ \|\ 0 < x < 1,\ 0 < y < 1\},\ ] ] see figure [ fig_face_map](2 ) . we denote by @xmath282 the torus which is the boundary of a regular neighborhood of @xmath231 . let @xmath283 be a primitive integral class . we set @xmath284 which consists of the parallel simple closed curves on @xmath282 . we define @xmath285 , @xmath286 and @xmath287 , @xmath288 in a similar way . [ lem_summary ] suppose that @xmath289 is a primitive integral class . * there is an orientation preserving homeomorphism @xmath290 which sends the minimal representative @xmath291 associated to @xmath292 to the oriented @xmath1-punctured disk bounded by the braid axis @xmath293 as in figure [ fig_braidhomeo](4 ) . thus the pseudo - anosov homeomorphism which represents the mapping class of @xmath294 corresponding to @xmath254 becomes the monodromy @xmath295 of the fibration associated to @xmath292 . * the boundary slope of @xmath296 ( resp . @xmath286 , @xmath288 ) is given by @xmath297 ( resp . @xmath298 , @xmath299 ) . * we have @xmath300 . the number of the boundary components of @xmath2 is computed as follows . @xmath301 where @xmath302 is defined by @xmath303 . * let @xmath304 be the monodromy of the fibration on @xmath0 associated to @xmath305 . then @xmath306 , and @xmath307 is conjugate to @xmath308 . * the dilatation @xmath309 is the largest root of @xmath310 * the ( un)stable foliation @xmath311 of @xmath55 has a property such that each component of @xmath312 , @xmath313 and @xmath314 has @xmath315 prongs , @xmath316 prongs and @xmath317 prongs respectively . moreover @xmath311 does not have singularities in the interior of @xmath318 . * @xmath54 is an orientable foliation if and only if @xmath270 and @xmath271 are even and @xmath272 is odd . the proof of ( 2 ) is easy . for the convenience of the proof of lemma [ lem_minrep ] , we prove the claim ( 1 ) . first of all , we observe that the link in figure [ fig_braidhomeo](2 ) is isotopic to @xmath319 given in figure [ fig_braidhomeo](1 ) . observe also that the link in figure [ fig_braidhomeo](3 ) is isotopic to the braided link @xmath42 given in figure [ fig_braidhomeo](4 ) . we use the link diagrams in ( 2 ) and ( 3 ) . we cut the twice punctured disk ( @xmath320 ) bounded by the component @xmath231 . let @xmath321 and @xmath322 be the resulting twice punctured disks obtained from @xmath235 . we reglue these twice punctured disks twisting either @xmath321 or @xmath323 by 360 degrees . then we obtain the link in ( 3 ) which is isotopic to @xmath42 . this implies that there exists an orientation preserving homeomorphism @xmath324 . then one can check that @xmath325 sends the minimal representative of @xmath292 to the desired @xmath1-punctured disk . chain link @xmath9 , ( 3)(4 ) braided link @xmath42.,width=508 ] lemma [ lem_summary](4 ) allows us to focus on only fibered classes @xmath305 such that @xmath326 for the proof of theorem [ thm_algorithm ] . we now introduce two bases @xmath327 and @xmath328 of @xmath329 to describe such fibered classes @xmath244 . if @xmath330 ( resp . @xmath331 ) , then we represent @xmath244 by the base @xmath327 ( resp . @xmath328 ) . let us define @xmath332 see figure [ fig_face_map](3)(4 ) , and define @xmath333 as follows . @xmath334 see figure [ fig_face_map](4 ) . for non - negative integers @xmath335 and @xmath136 , we define integral classes @xmath336 , @xmath337 to be @xmath338 where @xmath339 , @xmath340 , @xmath341 , and @xmath342 . the classes @xmath336 with @xmath343 are said to be _ non - degenerate_. other classes @xmath336 are said to be _ degenerate_. note that an integral class @xmath336 is fibered if and only if @xmath142 , @xmath141 are non - negative integers and @xmath344 is a positive integer , see ( [ equation_openface ] ) . if a fibered class @xmath305 satisfies @xmath326 and @xmath330 ( resp . @xmath326 and @xmath331 ) , then @xmath244 is written by @xmath345 ( resp . @xmath346 ) for some @xmath347 and @xmath348 . we use the notations @xmath349 and @xmath350 in the same manner as @xmath351 and @xmath352 appeared in lemma [ lem_summary ] . we denote by @xmath353 $ ] , the projection of @xmath4 to the fibered face @xmath258 . for simplicity , we write @xmath354= [ i , j , k]_{\pm}$ ] and @xmath355= [ j , k]_0 $ ] . clearly @xmath356_{\pm}\in \delta_{\pm}$ ] and @xmath357_0 \in \delta_0 $ ] . by claims ( 2)(3 ) in the following lemma , we find that coordinates @xmath336 are useful to study symmetries of the entropy function on @xmath0 . [ lem_summary2 ] let @xmath358 be a primitive integral class . * the dilatation @xmath350 is the largest root of @xmath359 in particular @xmath360 . * the integral class @xmath361 is a fibered class in @xmath362 , and the equality @xmath363 holds . in particular , @xmath364 . * two classes @xmath356_+$ ] and @xmath365_+$ ] have a line symmetry about @xmath366 , and @xmath356_+$ ] and @xmath356_-$ ] have a line symmetry about @xmath367 , see figure [ fig_face_map](5)(6 ) . the claim ( 1 ) holds by lemma [ lem_summary](5 ) . the first part of ( 2 ) follows from ( [ equation_openface ] ) . the second part of ( 2 ) is obvious from the claim ( 1 ) . the proof of ( 3 ) is easy to check . corollary 2.7 and remark 2.8 in @xcite . ) although all fibered classes @xmath336 and @xmath361 have the same thurston norm @xmath368 and same dilatation , the topological types of their fibers could be different in general . to see what the pseudo - anosovs @xmath369 and @xmath370 look like , we will see the curve complexes associated to @xmath336 and @xmath361 in section [ subsection_clique ] . and the fibered face @xmath258 . ( 2 ) @xmath371 \in \delta$ ] . ( 3 ) @xmath372 . ( 4 ) @xmath373 , @xmath374 . ( 5 ) projective classes of non - degenerate @xmath356_{\pm}$ ] and @xmath365_{\pm}$ ] . ( 6 ) projective classes of degenerate @xmath375_{\pm}$ ] and @xmath376_0= [ 0,i , k]_{\pm}$].,width=480 ] in section [ subsection_seed ] , we construct the pseudo - anosov homeomorphism on @xmath120 which represents the mapping class corresponding to the @xmath1-braid @xmath254 . it serves the monodromy @xmath295 of the fibration on @xmath0 associated to @xmath292 , and plays a rule as a pseudo - anosov homeomorphism @xmath19 in theorem [ thm_fried2 ] . we also construct a train track representative @xmath377 of @xmath378 $ ] . oertel used branched surfaces to describe thurston norm and to study fibers of fibrations on hyperbolic , fibered @xmath1-manifolds . for basic definitions and results on branched surfaces , see @xcite . in section [ subsection_minimal ] , we find minimal representatives of non - fibered classes @xmath379 , @xmath380 , @xmath381 , where @xmath382 and @xmath383 . in section [ subsection_branched ] , by using minimal representatives found in section [ subsection_minimal ] , we build two branched surfaces @xmath384 which carry fibers @xmath385 of fibrations associated to any fibered class @xmath336 . then in section [ subsection_train ] , we construct the train track @xmath386 which carries the unstable foliation @xmath387 of the pseudo - anosov monodromy @xmath388 of the fibration associated to @xmath336 . in section [ subsection_monodromies ] , we construct the the pseudo - anosov @xmath388 explicitly . in section [ subsection_ttmap ] , we give an explicit construction of the train track representative @xmath389 for @xmath390 $ ] . we also construct the directed graph @xmath391 induced by @xmath392 to indicate where each real edge of @xmath386 maps to under @xmath393 . in section [ subsection_clique ] , we give the curve complex @xmath394 of @xmath391 and compute its clique polynomial @xmath395 whose largest root equals the dilatation @xmath350 . let @xmath396 be the linear map induced by @xmath397 . since @xmath398 has eigenvalues @xmath399 with @xmath400 and @xmath401 , the linear map @xmath402 descends to the anosov diffeomorphism @xmath403 on the torus . figure [ fig_seed ] is an illustration of the image of the unit square with the origin ( on the left bottom corner ) under @xmath402 . ( left ) and its image @xmath404 ( right ) . ( the images of the isosceles right - angled triangles @xmath405 , @xmath406 , @xmath407 and @xmath408 under @xmath402 are the acute - angled triangles @xmath409 , @xmath410 , @xmath411 and @xmath412.),width=576 ] the linear map induced by @xmath413 defines a reflection @xmath414 . the quotient of @xmath415 by @xmath416 , denoted by @xmath417 , is homeomorphic to a sphere . ( @xmath417 is called a _ pillowcase _ because of its shape . ) we set the points @xmath418 , @xmath419 . let @xmath420 be the rectangle @xmath421 on @xmath422 , see figure [ fig_rectangle](1 ) . then @xmath417 is obtained from @xmath423 by identifying the three pairs of the oriented closed segments @xmath424 with @xmath425 , @xmath426 with @xmath427 , and @xmath428 with @xmath429 , see figure [ fig_rectangle ] . if we let @xmath430 be the composition of the projections @xmath431 and @xmath432 , then the differentiable structure of @xmath417 has the four singularities @xmath433 , @xmath434 , @xmath435 and @xmath436 which lie on the corners of @xmath417 . we set @xmath437 . . ( 2 ) @xmath41-punctured sphere @xmath438.,width=336 ] we have the identity @xmath439 , and this implies that @xmath440 induces a homeomorphism @xmath441 . clearly @xmath442 is invariant under @xmath443 . ( more concretely , @xmath444 for @xmath445 , @xmath446 and @xmath447 . ) we see that the homeomorphism @xmath443 on @xmath417 away from @xmath442 inherits the ( un)stable foliation of the anosov @xmath448 . therefore , @xmath440 induces a pseudo - anosov homeomorphism on @xmath449 , see figure [ fig_rectangle](2 ) . abusing the notation , we denote the pseudo - anosov on @xmath450 by the same notation @xmath443 . figure [ fig_papillow ] is an illustration of @xmath451 . figure [ fig_seed ] . ) let @xmath452 , @xmath453 , @xmath454 and @xmath455 be isosceles right - angled triangles whose vertices are punctures of @xmath120 , see figure [ fig_papillow](left ) . their images under @xmath443 , denoted by @xmath456 , @xmath457 , @xmath224 and @xmath458 , look as in figure [ fig_papillow](right ) . observe that the ( un)stable foliation of @xmath443 has a @xmath47-pronged singularity at each puncture . if we regard the puncture @xmath459 as the boundary of the @xmath1-punctured disk , then the mapping class @xmath460 $ ] is written by a @xmath1-braid on the disk . since @xmath398 is of the form @xmath461 we see that @xmath460 $ ] is represented by the @xmath1-braid @xmath462 ( reading the word from the right to the left ) . equivalently @xmath460 $ ] is written by @xmath463 , where @xmath464 denotes the mapping class which represents the positive half - twist about the segment @xmath465 , see figure [ fig_rectangle](2 ) . [ rem_homeo ] we have a natural homeomorphism @xmath466}$ ] . the cusp of the component @xmath467 ( resp . @xmath468 ) of the link @xmath469 maps to ( under @xmath470 ) the cusp corresponding to the orbit of @xmath459 ( resp . @xmath471 ) of the suspension flow ( see figure [ fig_3braid](2 ) ) . the cusp of the component @xmath293 maps to ( under @xmath470 ) the cusp corresponding to the orbit @xmath472 ( or @xmath473 ) . by lemma [ lem_summary](1 ) together with remark [ rem_homeo ] , we can regard @xmath451 as the monodromy @xmath295 of the fibration on @xmath0 associated to @xmath292 . we denote @xmath443 by @xmath474 . next we turn to a train track representative of @xmath378 $ ] . let @xmath475 be a train track on @xmath120 as in figure [ fig_seedabcd](1 ) . figures [ fig_seedabcd](1)(5 ) show that @xmath475 is invariant under @xmath476 . in fact , we have the image @xmath477 in figure [ fig_seedabcd](2 ) . ( for the illustration of @xmath477 , consider the image of the acute - angled triangle @xmath457 , see figure [ fig_papillow](right ) . ) the train track @xmath478 is isotopic to the one as in figure [ fig_seedabcd](5 ) which is carried by @xmath475 . as a result , we get the desired train track representative @xmath377 of @xmath476 whose incidence matrix of @xmath479 ( with respect to the real edges @xmath480 and @xmath481 ) is equal to @xmath482 . in the rest of the paper , we consider the magic manifold @xmath0 of the form @xmath483/ \sim$ ] , where @xmath32 identifies @xmath484 with @xmath485 for each @xmath486 . we investigate the suspension flow @xmath487 on @xmath488 . we choose the orientation of @xmath489 so that its normal direction coincides with the flow direction of @xmath487 . we fix the illustration of the @xmath41-punctured sphere @xmath489 as in figure [ fig_rectangle](2 ) , but we often omit the names of the punctures @xmath490 s . [ rem_real - edge ] loop edges of @xmath475 surrounding punctures @xmath471 , @xmath472 are @xmath473 are infinitesimal edges for @xmath377 . other edges @xmath480 and @xmath481 are real edges . each component of @xmath491 is a once punctured monogon . ( this comes from the fact that the ( un)stable foliation of @xmath443 has a @xmath47-pronged singularity at each puncture . ) in particular the component of @xmath491 containing the puncture @xmath459 has exactly one cusp . . , width=384 ] . ( 2 ) the image of @xmath475 under @xmath492 , where the image of the edge @xmath480 etc . is denote by @xmath493 etc . the image @xmath494 can be put in the tie neighborhood of @xmath475 under the isotopy , see ( 2)@xmath495(3)@xmath495(4)@xmath495(5 ) . ( from ( 4 ) @xmath495 ( 5 ) , the edges in ( 4 ) that are isotopic to the edges @xmath496 and @xmath497 in ( 2 ) cross over the segment @xmath220 ( see figure [ fig_rectangle](2 ) for @xmath220).,width=384 ] first we define several sets on @xmath500 ( see figure [ fig_parallel](1 ) ) . let @xmath501 and @xmath502 be the triangles @xmath503 and @xmath504 respectively . let @xmath505 and @xmath506 be the parallelograms @xmath507 and @xmath508 respectively . see figure [ fig_parallel](1 ) . note that @xmath509 where @xmath430 is the projection in section [ subsection_seed ] . let @xmath510 be the oriented closed segment @xmath511 . let @xmath512 and @xmath513 be the oriented closed segments @xmath514 and @xmath515 . , @xmath516 , @xmath517 , @xmath518 , @xmath519 , @xmath520 , @xmath521 . ( 2 ) @xmath522 . ( 3 ) @xmath523 , @xmath524.,width=480 ] . ( 2 ) @xmath525 . ( 3 ) @xmath526 . ( 4 ) @xmath527 . , width=432 ] let @xmath528 be the image @xmath529 removing all points in @xmath442 , i.e , @xmath530 . we define @xmath525 , @xmath526 , @xmath531 in the same manner . see figure [ fig_triangle ] . similarly , we let @xmath532 said differently , @xmath533 is obtained from @xmath534 by removing the end points . we define @xmath535 in the same manner . see figures [ fig_rectangle](2 ) and [ fig_parallel](2)(3 ) . we choose the orientations of @xmath536 , @xmath537 , @xmath538 which coincide with the ones induced by the fiber @xmath539 of the fibration associated to @xmath292 . [ rem_ukappa ] we have @xmath540 and @xmath541 , see figures [ fig_parallel](2 ) and [ fig_triangle](3)(4 ) ) . we have @xmath542 . ( we can check the equality by using figure [ fig_seed ] . ) this implies that @xmath543 in @xmath544 , and hence the flowband @xmath545 $ ] with respect to @xmath487 is defined . let @xmath546 , \ \mbox{and}\ \\ f_b & = & t_{b}^0 \cup [ i^0 , u^0 ] , \end{aligned}\ ] ] see figure [ fig_normone](1)(2 ) . observe that the both @xmath547 and @xmath548 are @xmath1-punctured spheres in @xmath544 . we find two more @xmath1-punctures spheres @xmath549 in @xmath544 . note that @xmath550 and @xmath551 . we have that @xmath552 and @xmath553 , because @xmath554 and @xmath555 . hence @xmath556 and @xmath557 in @xmath544 , and the flowbands @xmath558 $ ] and @xmath559 $ ] are defined . let @xmath560 \cup [ k^0 , w^0],\ \mbox{and}\ \\ f_{k_- } & = & p_{k_- } \cup [ j^0 , v^0 ] \cup [ k^0 , w^0 ] , \end{aligned}\ ] ] see figure [ fig_normone](3)(4 ) . we choose the orientations of @xmath547 , @xmath548 and @xmath549 so that they are extended by the orientations of @xmath561 , @xmath562 and @xmath563 . ( here we identify @xmath561 etc . with @xmath536 etc . ) [ lem_minrep ] the @xmath1-punctures spheres @xmath547 , @xmath548 , @xmath564 and @xmath565 are minimal representatives of @xmath379 , @xmath380 , @xmath566 and @xmath567 respectively . let @xmath568 , @xmath569 and @xmath570 be the meridians of the components @xmath231 , @xmath232 and @xmath233 of the @xmath1 chain link @xmath9 . we take oriented simple closed curves @xmath571 , @xmath572 and @xmath573 in @xmath574 as in figure [ fig_scc ] . we observe that the images of @xmath568 , @xmath569 and @xmath570 under the homeomorphism @xmath575 are @xmath571 , @xmath572 and @xmath573 respectively ( up to isotopy ) . now , we consider the intersections @xmath576 between the surface @xmath577 and either @xmath571 , @xmath572 or @xmath573 . we have @xmath578 , @xmath579 , and @xmath580 . these imply that @xmath548 is a minimal representative of @xmath380 . by cut and past construction of the union of surfaces @xmath581 , we obtain a surface which is homeomorphic to @xmath582 . since @xmath583 $ ] and @xmath584 $ ] , we conclude that @xmath547 is a minimal representative of @xmath379 . let us consider the intersections between @xmath565 and either @xmath571 , @xmath572 , or @xmath573 . we have @xmath585 , @xmath586 , and @xmath587 . thus we conclude that @xmath565 is a minimal representative of @xmath498 . then we see that @xmath564 is a minimal representative of @xmath499 , because @xmath584 $ ] and @xmath588 can be obtained from @xmath589 by cut and past construction . by lemma [ lem_minrep ] , it makes sense to denote @xmath590 , @xmath548 , @xmath591 and @xmath565 by @xmath235 , @xmath236 , @xmath592 and @xmath593 respectively . -punctured spheres embedded in @xmath544 with the suspension flow @xmath594 . the top and bottom surfaces are @xmath595 and @xmath596 respectively . they are identified by @xmath474 . two edges with the same kind ( red , green , yellow ) are identified in @xmath544 . the vertical arrows with the labeling @xmath597 indicate the flow directions . ( 1 ) @xmath598 . ( 2 ) @xmath599 . ( 3 ) @xmath600 . ( 4 ) @xmath601 . ( 5 ) @xmath602 for some @xmath603.,width=528 ] , @xmath573 and @xmath235 in @xmath544 . ( 2 ) @xmath571 , @xmath573 and @xmath236 in @xmath544.,width=240 ] in the end of this subsection , we introduce surfaces @xmath602 . we denote by @xmath602 , the oriented surface in @xmath544 which is obtained from @xmath236 by pushing @xmath604 along the flow direction for @xmath605 times , see figure [ fig_normone](5 ) . clearly @xmath606 $ ] . in the same manner we can define other surfaces @xmath607 and @xmath608 . we first construct the branched surface @xmath102 . we choose @xmath609 so that @xmath610 . we consider surfaces @xmath611 , @xmath612 , @xmath613 . we have @xmath614 $ ] , @xmath615 $ ] and @xmath616 $ ] . let @xmath617 the intersection of each pair of the three surfaces is as follows . 1 . @xmath618 , 2 . @xmath619 , 3 . ( see remark [ rem_ukappa ] and figure [ fig_normone](3)(5 ) . ) the local picture near each intersection looks as in figure [ fig_branchedsurface](1 ) . a common property of the pairs is that locally , a surface near the intersection is parallel to @xmath291 , and it intersects with the flowband which is the subset of the other surface ( in fact @xmath611 or @xmath621 ) . the branched surface @xmath102 can be obtained from @xmath622 by modifying each flowband of @xmath623 and @xmath624 as in figure [ fig_branchedsurface](2 ) so that the modifications agree with the orientations of the two surfaces . each point in the intersection of surfaces belongs to the _ branched locus _ of @xmath102 ( that is , the union of points of the branched surface none of whose neighborhood are manifolds ) . we build the branched surface @xmath103 in the same manner : take @xmath625 which represents @xmath626 instead of @xmath611 . let @xmath627 we have 1 . @xmath628 , 2 . @xmath629 , 3 . ( see remark [ rem_ukappa ] and figure [ fig_normone](4)(5 ) . ) the branched surface @xmath103 is obtained from @xmath631 by modifying each flowband of @xmath632 and @xmath624 in the similar manner as in the construction of @xmath102 . figure [ fig_identify](1 ) ( resp . figure [ fig_identify](3 ) ) illustrates three pieces ( bottom , middle , top ) for building @xmath102 ( resp . @xmath633 ) . in this figure , we have two kinds ( solid / broken ) of segments without arrows and two kinds ( solid / broken ) of segments with arrows . the broken segments without arrows are parts of the orbits of punctures @xmath471 , @xmath472 , @xmath473 and @xmath634 ( which are circles in the figure ) under the flow . in other words , they lie on the cusps of @xmath574 . we now explain our convention of figure [ fig_identify](1)(3 ) . firstly , there are some pairs of segments with the same labeling . ( exceptionally , the three segments in figure [ fig_identify](1 ) have the labeling l6 . ) the two segments with the same labeling mean that one of them is connected to the other with respect to the flow , see definition [ def_flowband ] . for the exceptional labeling l6 in figure [ fig_identify](1 ) , the bottom segment with the labeling l6 is connected to the middle segment with l6 , and the top segment with l6 is connected to the bottom segment with the same labeling . secondly , we also identify the segment having the labeling @xmath635 with the segment having the labeling @xmath636 . the resultant belongs to the branched locus . ( for example , the segment with the labeling @xmath637 and the two segments with the labeling @xmath638 are identified , and the resultant segment belongs to the branched locus . ) lastly , we can obtain the whole pictures of @xmath384 if we insert a suitable flowband between every two segments with the same labeling . for example , the bottom segment with the labeling @xmath638 is connected to the top segment with the same labeling @xmath638 . we insert a suitable flowband ( of the form @xmath639 $ ] ) between them . also the top segment with the labeling @xmath640 is connected to the bottom segment with the same labeling @xmath640 . thus we insert the flowband @xmath641= [ w^0 , w^{\delta}]$ ] between them . ( note that @xmath642 in @xmath544 . ) under the identification of segments and inserting suitable flowbands , each polygon bounded by the solid segments becomes a _ sector _ of the branched surface . in general , the _ sectors _ of the branched surface @xmath410 are the closures in @xmath410 of the components of @xmath643 ) . we turn to find surfaces carried by @xmath384 . to do this , given a fibered class @xmath336 ( hence @xmath644 , @xmath645 and @xmath348 ) , we assign these integers @xmath646 , @xmath647 and @xmath136 for the sectors of @xmath648 ( resp . @xmath649 ) as in figure [ fig_identify](2 ) ( resp . figure [ fig_identify](4 ) ) . this is a natural assignment , which we explain the reason now . we assign integers @xmath650 , @xmath645 and @xmath348 to @xmath611 , @xmath612 and @xmath651 ( resp . @xmath625 , @xmath612 and @xmath651 ) consisting of @xmath652 ( resp . @xmath653 ) . then we reconstruct @xmath648 ( resp . @xmath649 ) with the integers assigned . what we obtain is the assignment of integers in question . then the branched surface @xmath384 enjoys the branch equations of a particular type such that @xmath654 , @xmath655 and @xmath656 as in figure [ fig_branchedsurface_1d](1 ) . thus this assignment determines a surface @xmath657 which is carried by @xmath384 . ( see the illustration of figure [ fig_branchedsurface_1d](3 ) which shows the surface induced by some branch equation . ) said differently , @xmath658 is obtained from the union @xmath659 by cut and past construction of surfaces ( cf . figure [ fig_branchedsurface ] ) . therefore @xmath660 $ ] . . ( 2 ) side view of ( 1 ) in this figure . ( 3 ) side view of the surface induced by the branch equation . ( in this case @xmath661 , @xmath662.),width=384 ] . ( 2 ) assignment of sectors of @xmath102 . ( 3 ) three pieces for building @xmath103 . ( 4 ) assignment of sectors of @xmath103 . the types of arrows ( red , green , yellow ) are compatible with the ones illustrated in figure [ fig_normone].,width=480 ] [ lem_fiber ] the surface @xmath658 is the minimal representative of the fibered class @xmath336 . by definition , @xmath372 is a convex hull in @xmath258 containing @xmath381 ( @xmath663 $ ] , @xmath664 $ ] ) , @xmath665 $ ] , @xmath666 $ ] , see figure [ fig_face_map](3 ) . the fibered class @xmath336 is in the cone over @xmath372 , and the surface @xmath658 is built from @xmath667 ( which is the union of parallel copies of the minimal representatives @xmath668 , @xmath236 and @xmath291 ) by cut and past construction . thus @xmath658 must be the minimal representative of @xmath336 . [ rem_fiber ] by lemma [ lem_fiber ] , we write @xmath669 , and @xmath385 becomes a fiber of the fibration associated to @xmath336 . by theorem [ thm_fried2 ] , @xmath385 ( up to isotopy ) is transverse to the flow @xmath487 . the first return map @xmath388 with respect to @xmath487 becomes the pseudo - anosov monodromy of the fibration associated @xmath336 . we note that the unstable foliation @xmath670 of @xmath671 is carried by @xmath672 . let @xmath673 be the suspension of @xmath670 by @xmath474 which is the @xmath234-dimensional foliation of @xmath544 . we now construct the branched surfaces @xmath674 , each of which carries @xmath675 . let @xmath676 and @xmath677 be as in section [ subsection_branched ] , that is @xmath676 and @xmath677 are constants such that @xmath678 . hereafter we fix @xmath679 . we choose two families @xmath680 and @xmath681 of train tracks with the following properties . ( see figure [ fig_bsurface_latter ] , in which the time @xmath597 increases along arrows . ) 1 . 2 . @xmath683 for @xmath684 . @xmath685 is obtained from @xmath686 by folding edges of @xmath686 for each @xmath687 , or @xmath685 is isotopic to @xmath686 , 4 . @xmath688 for @xmath689 , and @xmath690 is given as in figure [ fig_bsurface_latter](3 ) , 5 . @xmath691 ( resp . @xmath692 ) is given as in figure [ fig_bsurface_latter](1 ) ( resp . ( 2 ) ) . in figure [ fig_bsurface_latter ] , @xmath693 { } \tau$ ] ( resp . @xmath694 { } \tau$ ] ) means that @xmath117 is obtained from @xmath125 by folding edges of @xmath125 , ( resp . @xmath117 is isotopic to @xmath125 ) . observe that non - loop edges of @xmath695 ( resp . @xmath696 ) do not intersect with @xmath523 and @xmath524 ( resp . with @xmath522 ) , see figure [ fig_bsurface_latter](4)(5 ) . the branched surfaces @xmath697 are defined to be @xmath698 where @xmath32 identifies @xmath699 and @xmath700 for @xmath701 . are indicated near the puncture @xmath473 . ) ( 1 ) @xmath702 . ( 2 ) @xmath703 . ( 3 ) @xmath704 . ( 4 ) @xmath705 . ( 5 ) @xmath696.,width=576 ] . ( 2 ) figure [ fig_identify](1)(3).),width=480 ] the condition ( 5 ) above makes the difference between @xmath707 and @xmath708 . the conditions ( 1)(4 ) ( without ( 5 ) ) allow us to construct a branched surface which carries @xmath675 . the reason we require ( 5 ) is that it is easy to extract a train track on @xmath385 from the intersection @xmath709 with the extra condition ( 5 ) , see lemma [ lem_pretraintrack](2 ) . [ rem_twiceonce ] the following analysis is used in the proof of lemma [ lem_pretraintrack](2 ) . it happens twice that an edge of some element in the subfamily @xmath710 is passing through the segment @xmath220 through the isotopy , see figure [ fig_bsurface_latter](2)(3 ) . on the other hand , the same thing happens once in the subfamily @xmath711 , see figure [ fig_bsurface_latter](1)(3 ) . in the same figure , @xmath41-punctured disks containing the track tracks in question are colored , and edges of these train tracks in question are made thick . since @xmath385 ( up to isotopy ) is transverse to the flow @xmath487 , we may assume that @xmath385 is transverse to @xmath712 . we let @xmath713 [ lem_pretraintrack ] 1 . the unstable foliation @xmath387 of the pseudo - anosov @xmath369 is carried by @xmath714 . each component of @xmath715 is either a bigon ( a disk with @xmath234 cusps ) or a once punctured disk ( i.e , annulus ) with @xmath344 cusps for some @xmath348 . \(1 ) by theorem [ thm_fried2 ] , we have @xmath716 . moreover its suspension @xmath717 by @xmath369 is isotopic to @xmath675 , see ( * ? ? ? * corollary 3.2 ) . since @xmath675 is carried by the branched surface @xmath674 , so is @xmath717 . this implies that @xmath387 is carried by @xmath718 . \(2 ) we recall that the suitable branch equations on @xmath384 induce @xmath385 . also recall that we need to insert some flowbands between the two segments with the same labeling to get the whole picture of @xmath384 . to view components of @xmath715 , let us consider @xmath719 on @xmath384 which is a @xmath47-branched manifold . we recall the definition of @xmath720 , see ( [ equation_pre - bpulus ] ) and ( [ equation_pre - bminus ] ) . the constant @xmath676 is chosen to be @xmath721 and we have the train tracks @xmath722 , @xmath696 , @xmath705 , see figure [ fig_bsurface_latter ] . then the ` patterns ' in figure [ fig_p_train](1)(2 ) are obtained from @xmath723 after folding or splitting all edges which appear on the flowbands . in fact the thick edge @xmath724 in figure [ fig_p_train](1 ) is the one ( resp . thick edges @xmath725 and @xmath726 in figure [ fig_p_train](2 ) are the ones ) by folding some edge ( resp . by splitting some edges ) which appear(s ) on the flowband between the segments with the labeling @xmath638 , see also remark [ rem_twiceonce ] . then we reconstruct the fibers @xmath385 obtained from the suitable branch equations on @xmath384 with the ` patterns ' in figure [ fig_p_train](1)(2 ) . that we get is @xmath714 ( up to folding and splitting the edges ) on @xmath385 . each edge of @xmath714 originates in some edge of the branched @xmath47-manifold @xmath723 . we denote some of the edges of @xmath727 ( resp . @xmath728 ) by @xmath480 , @xmath481 , @xmath168 , @xmath729 ( resp . @xmath480 , @xmath481 , @xmath168 , @xmath730 ) as in figure [ fig_p_train](1 ) ( resp . ( 2 ) ) . we can fold all edges of @xmath731 ( resp . @xmath732 ) which originate in @xmath724 ( resp . @xmath725 or @xmath726 ) into some edges . this means that these edges lie on the boundaries of some components of @xmath715 that are bigons . we fold all these edges as much as possible ( i.e , collapse bigons ) , and we consider complementary regions of the resulting @xmath47-branched manifold . the combinatorics from figures [ fig_identify](2)(4 ) and [ fig_p_train ] tell us that each component of the resulting @xmath47-branched manifold is a once punctured disk with @xmath136 cusps for some @xmath348 . let @xmath386 be the branched @xmath47-manifold obtained from @xmath714 by collapsing all bigons of @xmath714 . by lemma [ lem_pretraintrack ] , we immediately have : @xmath386 is a train track on @xmath385 which carries @xmath387 . first of all , we represent fibers @xmath385 more simpler as follows . we shrink each flowband of @xmath385 as much as possible along flow lines into some edge . ( note that this operation does not change the topological type of fibers . ) said differently , we simplify the branched surfaces @xmath384 as follows . shrink each flowband of the three pieces , see figure [ fig_identify](1 ) ( resp . ( 3 ) ) , as much as possible along flow lines into some edge . then the branch equations on @xmath384 induces the one on such a simplified branched surface , from which one gets a surface in question which is homeomorphic to @xmath385 . after shrinking flowbands , the resulting three pieces ( the bottom , middle , top pieces ) are : 1 . @xmath734-patch : the two acute - angled triangles sharing vertices @xmath471 and @xmath459 + ( resp . @xmath626-patch : the parallelogram ) , 2 . @xmath380-patch : the right - angled triangle , and 3 . @xmath292-patch : the rectangle . we call these pieces ( 1 ) _ @xmath734-patch _ _ @xmath626-patch _ ) , ( 2 ) _ @xmath380-patch _ and ( 3 ) _ @xmath292-patch_. see figure [ fig_map1_precise](1 ) ( resp . figure [ fig_map2_precise](1 ) ) for three kinds of patches . we often draw pairs of the two acute - angled triangles for the @xmath734-patch separately as in figure [ fig_map1_precise](1 ) , but they should share the two vertices @xmath471 and @xmath459 ( cf . figure [ fig_triangle](3 ) ) . we can get the simplified fibers @xmath735 ( resp . @xmath736 ) from @xmath142 parallel copies of @xmath734-patches ( resp . @xmath626-patches ) , @xmath141 parallel copies of @xmath380-patches and @xmath136 parallel copies of @xmath292-patches under suitable identifications of the boundaries of patches . if we have @xmath141 parallel copies of the same kind of patches , say @xmath380-patches , @xmath737 , we have @xmath738 for some @xmath739 , where @xmath740 . for the notation of @xmath741 , see the end of section [ subsection_minimal ] . if @xmath742 , then we call the @xmath380-patch @xmath743 _ the bottom ( @xmath380-)patch _ , and call @xmath744 _ the top ( @xmath380-)patch_. other @xmath380-patches are called the middle ( @xmath380-)patches . we define the top , middle , bottom for other patches similarly . we color each top of the three kinds of patches , see figure [ fig_map1_precise](1)(left column ) , figure [ fig_map2_precise](1)(left column ) . here we label @xmath745 for the top @xmath380-patch , and label @xmath746 and @xmath747 ( resp . @xmath746 ) for the top @xmath734-patch ( resp . @xmath626-patch ) in the same figure . we label @xmath452 , @xmath453 , @xmath454 and @xmath455 for isosceles right - angled triangles which lies on the top @xmath292-patch . the patches needed for building @xmath385 for the non - degenerate class @xmath336 are given as in the same figure . we can think three kinds of patches are in the cylinder @xmath748 . we have the flow direction in the cylinder from the ` bottom ' @xmath596 to the ` top ' @xmath595 . the types of arrows ( red , green , yellow ) in the figure are compatible with the ones illustrated in figures [ fig_normone ] and [ fig_identify ] . among the patches with the same kind ( @xmath381-patches , @xmath380-patches or @xmath292-patches ) , the way to label parallel segments with the same kind of arrow is that the number for the labeling increases ( cyclically ) along the flow direction . we often omit to label segments which lie on the middle patches . to get the fiber @xmath385 , we identify the two segments with the same kind of arrow and with the same labeling ( same number ) by using the flow @xmath487 . in the right column of ( 1 ) in the same figure , the labeling of segments on patches are the same as the one given in the left column . let us turn to construct @xmath749 explicitly . it is enough to describe where each patch maps to . since the desired monodromy @xmath369 is the first return map on @xmath385 with respect to @xmath487 , we see the followings . all patches but the top of each kind of patches map to the next above patch ( of the same kind ) along the flow direction . thus the monodromy @xmath369 restricted to these patches is just a shift map . on the other hand , each top patches map to some bottom patches ( possibly with different kinds ) , see figure [ fig_map1_precise](1 ) for @xmath750 and see figure [ fig_map2_precise](1 ) for @xmath751 , where @xmath752 are the images of @xmath753 under @xmath369 . more precisely , we can get the image of the top @xmath380-patch under @xmath369 when we push the fiber @xmath385 along the flow direction and see how this top patch hits to the bottom @xmath292-patch . similarly , one can get the image of the top @xmath734-patch ( resp . top @xmath626-patch ) under @xmath754 ( resp . @xmath755 ) if we see how this top patch hit to the both bottom @xmath380-patch and bottom @xmath292-patch . to get the images of the isosceles right - angled triangles @xmath452 , @xmath453 , @xmath454 and @xmath455 which lies on the top @xmath292-patch , we first consider the acute - angled triangles @xmath456 , @xmath457 , @xmath224 and @xmath458 which lie on @xmath596 as in figure [ fig_papillow](right ) . then investigate how these acute - angled triangles hit bottoms patches , when we push them along the flow direction . the monodromies of the fibrations associated to the degenerated classes can be constructed similarly . as an example , we deal with the degenerated classes @xmath756 s , see figure [ fig_map3_precise ] . [ rem_delta0 ] suppose that @xmath757 is primitive ( i.e , @xmath758 ) . then the fiber @xmath759 is connected , and it has genus @xmath760 , see @xcite . many pseudo - anosovs with small dilatations defined on the surfaces of genus @xmath760 are contained in the family of fibered classes @xmath757 s , see examples [ ex_braid1 ] , [ ex_braid2 ] and ( * ? ? ? * section 4.1 ) . by using the artin generators of the braid groups , the words which represent @xmath761 s are given in ( * theorem 3.4 ) . they are quite simple words . for non - degenerate @xmath762 . ( 2 ) ( left ) bottom edges @xmath763 , @xmath764 , @xmath765 and @xmath766 which lie on three bottom patches . ( right ) images of top edges @xmath767 , @xmath768 , @xmath769 and @xmath770 under @xmath771 which lie on three bottom patches . we denote by @xmath772 etc . , the image @xmath773 etc.,width=528 ] for non - degenerate @xmath774 . ( 2 ) ( left ) bottom edges @xmath775 , @xmath764 , @xmath765 and @xmath766 which lie on three bottom patches . ( right ) images of top edges @xmath776 , @xmath768 , @xmath769 and @xmath770 under @xmath771 which lie on three bottom patches.,width=480 ] for degenerate @xmath777.,width=336 ] for @xmath778 . ( 1)(2)(3 ) degenerate cases . ( 4 ) non - degenerate case . ( 1 ) @xmath779 with @xmath780 . ( 2 ) @xmath781 with @xmath782 . ( 3 ) @xmath756 with @xmath783 . ( 4 ) @xmath784 with @xmath343.,width=576 ] for @xmath785 . ( 1)(3 ) degenerate cases . ( 2)(4 ) non - generate cases . ( 1 ) @xmath786 with @xmath787 . ( 2 ) @xmath774 with @xmath787 . ( 3 ) @xmath786 with @xmath788 . ( 4 ) @xmath785 with @xmath789 . for ( 3)(4 ) , @xmath146 is an integer such that @xmath790.,width=556 ] in the following lemma , we use the metrized , directed graph @xmath391 given in figures [ fig_digraph_p ] and [ fig_digraph_n ] , where each edge with no labeling means that its length is equal to @xmath47 , and all edges with labeling are made thick in the figures . [ lem_invariant ] 1 . the train track @xmath386 is invariant under @xmath792 $ ] . if we let @xmath392 be the train track representative of @xmath791 , then its incidence matrix is perron - frobenius . the directed graph @xmath391 is the one induced by @xmath392 , once we prove that @xmath386 is invariant under @xmath791 , the claim that the incidence matrix of @xmath791 is perron - frobenius follows from theorem [ thm_pape ] . by construction of @xmath386 , each edge of @xmath386 originates in some edge of the intersection @xmath723 ( see figure [ fig_p_train ] ) . we only label edges of @xmath793 ( resp . @xmath794 ) which originate in the edges @xmath795 and @xmath729 ( resp . @xmath795 and @xmath730 ) , since it turns out that the images of other edges under @xmath369 are eventually periodic up to isotopy . we first explain the way to label the edges of @xmath386 , which is similar to the one to label segments of patches given in section [ subsection_monodromies ] . we label parallel edges of @xmath386 ( which originate in the same edge of @xmath723 ) so that the number for the labeling increases along the flow direction . for example , see figures [ fig_ex_lt][fig_ex_tsai ] . we only prove the claims ( 1 ) and ( 2 ) for non - degenerate classes @xmath336 . ( the proofs for degenerate classes are similar . ) let us describe the image of an edges @xmath133 with labeling under @xmath369 . such an edge @xmath133 lies on some patch for building @xmath385 . suppose that @xmath133 lies on a patch , say @xmath796 which is not a top patch . then @xmath133 maps ( under @xmath369 ) to the next above edge @xmath797 which lies on the same kind of patch @xmath798 as @xmath799 . clearly , @xmath133 and @xmath797 originate in the same edge of @xmath723 . suppose that an edge @xmath800 of @xmath386 lies on some top patch . we call @xmath800 the _ top edge_. if an edge @xmath801 of @xmath386 lies on some bottom patch , then we call @xmath801 the _ bottom edge_. we first consider the non - degenerate class @xmath345 . let us consider the images @xmath802 s for all top edges @xmath800 s . then we can put @xmath802 s in the tie neighborhood of @xmath803 which are transverse to the ties up to isotopy , where the support of the isotopy can be taken on the neighborhood of the three bottom patches , see figure [ fig_map1_precise](2 ) . thus @xmath803 is invariant under @xmath804 . in fact we can get the edge path @xmath805 from figure [ fig_map1_precise](2 ) : the left of ( 2 ) of the same figure illustrates the union of all bottom edges of @xmath793 which lie on the union of all bottom patches . the right of ( 2 ) of the same figure shows the image of all top edges under @xmath754 . one can find from the right in ( 2 ) of the same figure that @xmath806 and @xmath807 pass through the three edges @xmath764 , @xmath765 , @xmath763 , and the two edges @xmath764 , @xmath765 respectively . the edge path @xmath808 passes through the two edges @xmath766 , @xmath763 . note that by remark [ rem_real - edge ] , we see that edges of @xmath793 which originate in the edges @xmath795 and @xmath729 are real edges for @xmath809 . others are infinitesimal edges . we explain a structure of the directed graph in figure [ fig_digraph_p](4 ) . when @xmath133 is an edge with some label ( which are made thick in the figure ) , then the end points of @xmath133 are the vertices having the same origin ( @xmath480 , @xmath481 , @xmath168 or @xmath729 ) on @xmath102 . suppose that @xmath133 is an edge whose length equals @xmath810 , and suppose that @xmath133 has end points @xmath765 and @xmath769 having the same origin @xmath480 . then the edge @xmath133 with labeling @xmath810 corresponds to the following edge path with length @xmath810 : @xmath811 in particular all vertices between @xmath765 and @xmath769 have the same origin @xmath480 on @xmath102 . putting these things together , we can check that metrized , directed graph @xmath812 in figure [ fig_digraph_p](4 ) is the one induced by @xmath813 . we turn to the non - degenerate class @xmath346 . in the same manner as in the class @xmath345 , we can see that @xmath794 is invariant under @xmath814 . the edges of @xmath794 which originate in the edges @xmath795 and @xmath730 are real edges for the train track representative @xmath815 of @xmath814 . others are infinitesimal edges . a hint to obtain @xmath816 is given in figure [ fig_map2_precise](2 ) . the left of ( 2 ) in the same figure illustrates the union of all bottom edges of @xmath794 which lie on the union of all bottom patches . the right of ( 2 ) in the same figure shows the image of all top edges under @xmath755 . the images under @xmath815 of all top edges but @xmath768 are edge paths written by bottom edges . for example , @xmath817 is an edge path which passes through @xmath764 and @xmath765 . on the other hand , to put @xmath818 in the tie neighborhood of @xmath819 one needs to make @xmath818 ( up to isotopy ) across the segment @xmath220 . an analysis to identify boundaries of patches for building @xmath736 enables us to get the image @xmath820 . we can verify that the directed graph @xmath821 given in figure [ fig_digraph_n](2 ) ( resp . ( 4 ) ) is the one induced by @xmath815 if @xmath822 ( resp . @xmath789 ) . the edge path @xmath820 has length @xmath823 , where @xmath824 is an integer such that @xmath790 . metrized and directed graphs @xmath391 for non - degenerate classes can recover ones for degenerate classes . see figures [ fig_p_train](1)(2)(3 ) and [ fig_digraph_p](1)(3 ) . to see this , consider @xmath812 for the non - degenerate class @xmath345 . we have the edge @xmath825 and the edge path from @xmath763 to @xmath765 for @xmath812 , see figure [ fig_digraph_p](4 ) . connecting these edge paths , we have the edge path from @xmath769 to @xmath765 with length @xmath826 , see figure [ fig_digraph_p](4 ) . this determines _ the edge _ @xmath827 with length @xmath47 in the degenerate class @xmath828 by ` eliminating ' vertices @xmath829 , see figure [ fig_digraph_p](3 ) . another example is this . we have the edge @xmath830 and the edge path from @xmath763 to @xmath765 for the non - degenerate class @xmath345 , see figure [ fig_digraph_p](4 ) . they determine _ the edge _ @xmath831 with length @xmath47 for the degenerate class @xmath828 , see figure [ fig_digraph_p](3 ) . we define some graphs . let @xmath832 denote the complete bipartite graph with @xmath135 and @xmath70 vertices . we denote by @xmath833 , the disjoint union of @xmath832 and the graph with two vertices and with no edges . the metrized , directed graphs @xmath812 s and @xmath821 s were given in section [ subsection_ttmap ] . here we exhibit their curve complexes @xmath834 s in figure [ fig_wgraph_p ] and @xmath835 s in figure [ fig_wgraph_n ] . in the next proposition , we give the clique polynomial @xmath836 for the computation of @xmath350 . we find that it is equal to the polynomial @xmath837 in lemma [ lem_summary2](1 ) . [ prop_clique ] the clique polynomial @xmath395 of the curve complex @xmath394 is the following reciprocal polynomial @xmath838 in particular , the dilatation @xmath839 is the largest root of @xmath840 . it is straightforward to compute the clique polynomial @xmath395 . the dilatation @xmath350 equals the growth rate of @xmath391 . by theorem [ thm_mcmullen ] , @xmath841 is the smallest root of @xmath395 . note that @xmath395 is a reciprocal polynomial , i.e , @xmath842 . thus the largest root of @xmath395 equals @xmath350 . for @xmath843 . ( 1 ) @xmath844 if @xmath781 including the case @xmath845 and @xmath780 , see figure [ fig_digraph_p](1)(2 ) . ( 2 ) @xmath846 if @xmath784 including the case @xmath845 , @xmath847 and @xmath780 , see figure [ fig_digraph_p](3)(4).,width=384 ] for @xmath848 . ( 1 ) @xmath849 if @xmath785 with @xmath787 including the case @xmath850 , see figure [ fig_digraph_n](1)(2 ) . ( 2 ) @xmath851 if @xmath785 with @xmath789 including the case @xmath850 , see figure [ fig_digraph_n](3)(4 ) . here @xmath146 is an integer such that @xmath790 . the weights of @xmath852 vertices of the subgraph @xmath853 surrounding the centered vertex are given by @xmath854 , @xmath855 clockwise.,width=384 ] let @xmath4 be a fibered class of @xmath0 . by a symmetry of thurston norm ball @xmath245 and by lemma [ lem_summary](4 ) , we may suppose that @xmath4 is of the form @xmath856 . suppose that @xmath857 . in this case , we have constructed @xmath51 and @xmath101 explicitly in section [ subsection_seed ] . let us consider other fibered classes @xmath858 s . the explicit construction of the monodromy @xmath51 of the fibration associated to @xmath4 is given in section [ subsection_monodromies ] . if @xmath4 is primitive , then the explicit construction of the desired train track representative @xmath859 of @xmath104 together with the induced directed graph @xmath114 of @xmath860 ( figures [ fig_digraph_p ] and [ fig_digraph_n ] ) is given in section [ subsection_ttmap ] . recall that @xmath285 ( resp . @xmath282 , @xmath287 ) is a torus which is the boundary of a regular neighborhood of the component @xmath232 ( resp . @xmath231 , @xmath233 ) of the @xmath1 chain link @xmath9 . recall that @xmath105 is the manifold obtained from @xmath0 by dehn filling the cusp along the slope @xmath861 . it is known by @xcite that @xmath105 is hyperbolic unless @xmath862 . the manifolds @xmath107 and @xmath108 are the exterior of the whitehead sister link ( i.e , @xmath863-pretzel link ) and the exterior of the @xmath1-braided link @xmath864 ( or @xmath865 link in rolfsen s table ) respectively . also @xmath109 is the exterior of the whitehead link ( see @xcite for example ) . these dehn fillings @xmath107 , @xmath108 and @xmath866 play an important rule to study pseudo - anosovs with small dilatations ( see @xcite ) , which we recall quickly in section [ subsection_infinite ] . first of all , we consider the relation of fibered classes between @xmath0 and dehn fillings @xmath105 . we suppose that @xmath105 is the manifold obtained from @xmath0 by dehn filling the cusp specified by @xmath285 along the slope @xmath867 . by using lemma [ lem_summary](2 ) ( which says that the boundary slope of @xmath286 equals @xmath868 ) , we have the following : there exists a natural injection @xmath869 whose image equals @xmath870 , where @xmath871 see ( * ? ? ? * proposition 2.11 ) . we choose @xmath872 with @xmath873 , and let @xmath874 be a fibered class in @xmath875 . then @xmath876 is also a fibered class of @xmath105 . this is because the monodromy @xmath51 of the fibration on @xmath0 associated to @xmath4 extends to the monodromy @xmath877 of the fibration on @xmath105 associated to @xmath878 by capping each boundary component of @xmath286 with the disk . if the ( un)stable foliation @xmath54 of @xmath55 has a property such that each component of @xmath286 has no @xmath47 prong , then @xmath54 extends canonically to the ( un)stable foliation @xmath879 of @xmath877 , and hence @xmath877 becomes a pseudo - anosov homeomorphism with the same dilatation as @xmath55 . we sometimes denote @xmath105 by @xmath880 when we need to specify the cusp which is filled . by using this notation , we may write @xmath881 . similarly , when @xmath105 is the manifold obtained from @xmath0 by dehn filling the cusp specified by another torus @xmath287 along the slope @xmath168 , we have a natural injection , @xmath882 whose image @xmath883 is given by @xmath884 in the following example , we consider some fibered classes @xmath4 and compute their dilatations @xmath166 . we also exhibit @xmath885 , its image @xmath886 putting into the tie neighborhood @xmath887 ( so that it is transverse to the ties ) and the directed graph @xmath888 . 1 . if @xmath889 , then @xmath890 is the largest root of @xmath891 by lemma [ lem_summary](3 ) , the topological type of @xmath2 is @xmath892 . by lemma [ lem_summary](7 ) , @xmath54 is orientable . this implies that the monodromy @xmath51 of the fibration associated to @xmath4 extends to the pseudo - anosov @xmath893 on the closed surface of genus @xmath234 with orientable invariant foliations by capping each boundary component of @xmath894 with the disk . the dilatation of @xmath895 is same as that of @xmath55 , that is @xmath166 . the minimal dilatations @xmath81 and @xmath896 are computed in @xcite and @xcite respectively , and @xmath897 holds . in fact , @xmath81 is the largest root of @xmath898 , and hence we have @xmath899 . thus @xmath895 is a minimizer of @xmath897 . see figure [ fig_6_2_1 ] . 2 . if @xmath900 , then @xmath901 is the largest root of @xmath902 see figure [ fig_1_2_-3 ] . 3 . if @xmath903 , then @xmath904 is the largest root of @xmath905 the fiber @xmath2 is homeomorphic to @xmath906 . on the other hand , @xmath907 , and @xmath54 has a property such that each component of @xmath908 has no @xmath47 prong . hence by capping either @xmath296 or @xmath909 with the disk , we get the pseudo - anosov @xmath910 on the @xmath911-punctured sphere with the same dilatation as @xmath166 . since @xmath877 fixes the boundary component ( @xmath296 or @xmath909 ) of @xmath2 , it defines the pseudo - anosov on @xmath912 . lanneau and thiffeault computed @xmath913 in @xcite . we see that @xmath166 is equal to @xmath913 , and hence @xmath877 is a minimizer of @xmath913 . see figure [ fig_3_5_0 ] . we exhibit several infinite sequences of fibered classes of @xmath0 , each of which corresponds to a subsequence of pseudo - anosovs to give the upper bounds ( u1 ) , @xmath99 , ( u6 ) except ( u3 ) in table [ table_section ] . each sequence lies on the section either @xmath870 for @xmath917 or @xmath918 for @xmath919 . a sequence of fibered classes in @xmath920 ( resp . @xmath921 , @xmath922 ) defines the sequence of fibered classes of the hyperbolic dehn filling @xmath107 ( resp . @xmath109 , @xmath108 ) . we indicate where each sequence sits on the fibered face @xmath258 , see figure [ fig_candidatepl ] . the purpose of the following examples is to construct the invariant train track @xmath923 , the induced directed graph @xmath114 and curve complex @xmath115 associated to each fibered class @xmath4 in each infinite sequence concretely and to see a structure of the monodromy @xmath55 of the fibration associated to @xmath4 via the curve complex @xmath115 . we will see that the topological type of @xmath115 is fixed and it is either @xmath924 or @xmath925 for each infinite sequence . [ ex_lt ] suppose that @xmath930 @xmath931 . take a sequence of primitive fibered classes @xmath932 ( the fibered class in example [ example_ex](1 ) is equal to @xmath933 in the present example . ) observe that the projections @xmath934 $ ] s go to @xmath935 \in int(\delta)$ ] as @xmath7 goes to @xmath173 , see figure [ fig_candidatepl](1 ) . by lemma [ lem_summary](3)(7 ) , it was shown in @xcite that the fiber @xmath936 has genus @xmath7 , and @xmath937 is orientable . the dilatation @xmath938 is equal to the largest root of @xmath939 since @xmath937 is orientable , the monodromy @xmath940 of the fibration associated to @xmath941 extends canonically to the pseudo - anosov homeomorphism @xmath942 on the closed surface @xmath64 of genus @xmath7 with the dilatation @xmath938 . we now introduce the _ lanneau - thiffeault polynomial _ @xmath943 then @xmath944 is a factor of @xmath945 . if we set @xmath946 to be the largest root of @xmath944 , then we have @xmath947 . it is known that @xmath948 @xcite , @xmath949 @xcite and @xmath950 @xcite . lanneau and thiffeault asked in @xcite whether the equality @xmath951 holds for even @xmath7 . the existence of pseudo - anosovs @xmath942 for @xmath930 @xmath931 in the present example were discovered by hironaka in @xcite , see also remark [ rem_hironaka - san ] . the sequence of pseudo - anosovs @xmath952 can be used for a subsequence to prove both upper bounds ( u1 ) and ( u2 ) in table [ table_section ] . in fact we have @xmath953 [ ex_ori79 ] an explicit value of @xmath8 is known . the lower bound of @xmath8 was given in @xcite and an example which realizes this lower bound was found in @xcite and @xcite independently . however an explicit construction of a minimizer of @xmath8 was not given in @xcite . here we construct such a minimizer . suppose that @xmath954 . following @xcite , we take a sequence of primitive fibered classes @xmath955 then @xmath936 has genus @xmath7 ( @xcite ) and the projections @xmath934 $ ] s go to @xmath956 \in int(\delta)$ ] as @xmath7 goes to @xmath173 , see figure [ fig_candidatepl](2 ) . we have @xmath957 and the dilatation @xmath958 is equal to the largest root of this polynomial . in particular @xmath958 is equal to the largest root of the second factor of @xmath959 . lemma [ lem_summary](7 ) ensures that the monodromy @xmath940 of the fibration associated to @xmath941 extends to the pseudo - anosov @xmath960 with orientable invariant foliations . it is known that @xmath8 is equal to the largest root of the second factor of @xmath961 as above . hence @xmath962 , and the extension @xmath963 of @xmath964 is a minimizer of @xmath8 . the existence of the sequence of pseudo - anosovs @xmath952 was discovered in @xcite . we have @xmath965 see the upper bounds ( u1 ) and ( u2 ) . [ ex_ori15 ] suppose that @xmath966 . following @xcite , we take a sequence of primitive fibered classes @xmath967 then the fiber @xmath968 has genus @xmath7 ( @xcite ) , and the dilatation @xmath969 is equal to the largest root of @xmath970 observe that the projections @xmath934 $ ] s go to @xmath971 \in int(\delta)$ ] as @xmath7 goes to @xmath173 , see figure [ fig_candidatepl](2 ) . we see that the monodromy @xmath940 of the fibration associated to @xmath941 extends to the pseudo - anosov @xmath972 on @xmath64 with orientable invariant foliations . the dilatation of @xmath972 is same as the dilatation of @xmath973 , that is @xmath969 . the equality @xmath974 holds , see @xcite . the sequence of pseudo - anosovs @xmath952 satisfies @xmath965 see the upper bounds ( u1 ) and ( u2 ) . [ ex_whitehead ] following @xcite , we take a sequence of primitive fibered classes @xmath975 the dilatation @xmath976 is equal to the largest root of @xmath977 the projections of @xmath978 s go to @xmath979 \in int(\delta)$ ] as @xmath70 goes to @xmath173 , see figure [ fig_candidatepl](3 ) . we find that @xmath978 defines a fibered class @xmath980 , see the beginning of section [ section_catalogue ] . by using lemma [ lem_summary](6 ) , we see that the monodromy of the fibration on @xmath109 associated to @xmath981 has the dilatation @xmath976 . observe that the topological type of the fiber ( i.e , minimal representative ) of @xmath982 is homeomorphic to @xmath983 . the sequence of the monodromies @xmath984 of fibrations associated to @xmath982 s on the whitehead link exterior @xmath985 can be used for a subsequence to prove ( u5 ) . in fact , we have @xmath986 [ ex_braid1 ] for @xmath987 , consider a sequence of primitive fibered classes @xmath988 which is studied in @xcite . we see that @xmath989 is homeomorphic to @xmath990 . the projection of @xmath978 goes to @xmath991 \in int(\delta)$ ] as @xmath70 goes to @xmath173 , see figure [ fig_candidatepl](4 ) . the dilatation @xmath992 equals the largest root of @xmath993 by using the same arguments as in example [ example_ex](3 ) , we see that the monodromy @xmath994 of the fibration associated to @xmath978 defines the pseudo - anosov @xmath995 with the same dilatation @xmath992 . such a pseudo - anosov homeomorphism @xmath995 is studied in @xcite . it is known that @xmath996 , see @xcite and @xmath997 , see @xcite . the sequence of pseudo - anosovs @xmath998 can be used for a subsequence to prove ( u4 ) : @xmath999 [ ex_braid2 ] for @xmath1000 , let us consider a sequence of fibered classes @xmath1001 which is studied in @xcite . ( the fibered class in example [ example_ex](3 ) is equal to @xmath933 in the present example . ) we see that @xmath989 is homeomorphic to @xmath1002 . the projection of @xmath978 goes to @xmath991 \in int(\delta)$ ] as @xmath70 goes to @xmath173 . the dilatation @xmath1003 equals the largest root of @xmath1004 by using the same arguments as in example [ example_ex](3 ) , we see that the monodromy @xmath1005 of the fibration associated to @xmath978 defines the pseudo - anosov @xmath1006 with the dilatation @xmath1003 . as we have seen in example [ example_ex](3 ) that @xmath1007 holds . such a pseudo - anosov homeomorphism @xmath1006 is also studied in @xcite . the sequence of pseudo - anosovs @xmath1008 can be used for a subsequence to prove ( u4 ) : @xmath1009 [ ex_tsai ] throughout this example , we fix @xmath1010 . the following sequence of fibered classes is used for the proof of theorem [ thm_boundary ] , see @xcite : @xmath1011 the projections @xmath1012 $ ] s go to @xmath1013 \in \partial \delta$ ] as @xmath480 goes to @xmath173 , see figure [ fig_candidatepl](5 ) . it is not hard to see that @xmath1014 is primitive if and only if @xmath98 and @xmath1015 are relatively prime . if @xmath1014 is primitive , then the fiber @xmath1016 is homeomorphic to @xmath1017 , and we have @xmath1018 there exists a sequence of primitive fibered classes @xmath1019 with @xmath1020 when @xmath1021 , where @xmath1022 depends on @xmath7 . ( for example , take @xmath1023 . ) this means that @xmath1024 has genus @xmath7 and the number of the boundary components of @xmath1024 ( which is equal to @xmath1025 ) goes to @xmath173 as @xmath142 does . as we have seen above , the projection of such a class @xmath1026 goes to the same point @xmath1013 \in \partial \delta$ ] ( which does not depend on @xmath7 ) as @xmath1022 goes to @xmath173 . d. fried , _ fibrations over @xmath1045 with pseudo - anosov monodromy _ , expos 14 in ` travaux de thurston sur les surfaces ' by a. fathi , f. laudenbach and v. poenaru , astrisque , 66 - 67 , socit mathmatique de france , paris ( 1979 ) , 251 - 266 . e. hironaka , _ small dilatation pseudo - anosov mapping classes and short circuits on train track automata _ , mittag - leffler institute preprint collection , workshop on growth and mahler measures in geometry and topology ( 2013 ) , 17 - 41 . n. v. ivanov , _ stretching factors of pseudo - anosov homeomorphisms _ , journal of soviet mathematics , 52 ( 1990 ) , 28192822 , which is translated from zap . nauchu . sem . otdel . mat . inst . steklov . ( lomi ) , 167 ( 1988 ) , 111 - 116 . e. kin , _ notes on pseudo - anosovs with small dilatations coming from the magic 3-manifold _ , representation spaces , twisted topological invariants and geometric structures of 3-manifolds , rims kokyuroku 1836 ( 2013 ) , 45 - 64 . available at ` http://www.kurims.kyoto-u.ac.jp/~kyodo/kokyuroku/contents/pdf/1836-05.pdf ` s. matsumoto , _ topological entropy and thurston s norm of atoroidal surface bundles over the circle _ , journal of the faculty of science , university of tokyo , section ia . mathematics 34 ( 1987 ) , 763 - 778 .	we study the magic manifold @xmath0 which is a hyperbolic and fibered @xmath1-manifold . we give an explicit construction of a fiber @xmath2 and its monodromy @xmath3 of the fibration associated to each fibered class @xmath4 of @xmath0 . let @xmath5 ( resp . @xmath6 ) be the minimal dilatation of pseudo - anosovs ( resp . pseudo - anosovs with orientable invariant foliations ) defined on an orientable closed surface of genus @xmath7 . as a consequence of our result , we obtain the first explicit construction of the following pseudo - anosovs ; a minimizer of @xmath8 and conjectural minimizers of @xmath5 for large @xmath7 .
one of the first successes of quantum mechanics was the explanation of @xmath0-particle tunneling through barrier . about 20 years after the first experimental observation of the @xmath0 radioactivity @xcite , the theoretical explanation in terms of quantum mechanical barrier penetration @xcite was provided which was , at least , qualitatively successful . since then the @xmath0 decay half life measurements with constantly improving techniques and the quest for theoretical explanation of their absolute magnitudes have continued unabated . the attractive features of @xmath0 radiactivity are that the @xmath0 particles can be detected rather easily under favourable conditions such as high efficiency , low background and good energy resolution . about ten years ago , theoretical estimates of lifetimes for the @xmath0 and various exotic decays have been provided in detail by the analytical superasymmetric fission model ( asafm ) @xcite with reasonable success . this was followed by the cluster model ( cm ) calculations for the @xmath0 decay half lives @xcite for even more comprehensive database with similar success . but the nuclear interaction potentials used by both the theoretical approaches described above are phenomenological in nature and designed to fit the experimental data . the asafm uses a parabolic potential approximation for the nuclear interaction potential , which is a rather unusual @xmath0-nucleus interaction potential . the cm uses a strange cos - hyperbolic form for the nuclear interaction potential @xcite . nevertheless , they provide a benchmark against which more microscopically based treatments can be judged . in the present work , the nuclear potentials have been obtained microscopically by double folding the @xmath0 and daughter nuclei density distributions with a realistic m3y effective interaction . this is the ideal procedure of obtaining nuclear interaction energy for the @xmath0-nucleus interaction . any liquid drop like properties such as surface energy are basically macroscopic manifestation of microscopic phenomena . a double folding potential obtained using m3y effective interaction is more appropriate because of its microscopic nature . a potential energy surface is inherently embedded in this description . moreover , the use of global microscopic nuclear potentials for a wide range of @xmath0-nucleus interaction is also theoretically a very thorough approach . the double folding potential has then been utilised within the superasymmetric fission model description . this superasymmetric fission model using microscopic potentials has been used to provide estimates of @xmath0 decay half lives for a wide range of @xmath0-emitters . the semirealistic explicit density dependence on the m3y effective interaction has also been employed to study the effects of density dependence on the @xmath0 decay lifetimes . the penetrability of the pre - scission part of the potential barrier provides the @xmath0 cluster preformation probability within a superasymmetric fission model @xcite . the microscopic nuclear potentials @xmath1 have been obtained by double folding in the densities of the fragments @xmath0 and daughter nuclei with the finite range realistic m3y effective interacion as @xmath2 d^3r_1 d^3r_2 \label{seqn1}\ ] ] where the density distribution function @xmath3 for the @xmath0 particle has the gaussian form @xmath4 whose volume integral is equal to @xmath5 , the mass number of @xmath0-particle . the density distribution function @xmath6 used for the residual cluster , the daughter nucleus , has been chosen to be of the spherically symmetric form given by @xmath7 \label{seqn3}\ ] ] where @xmath8 and the value of @xmath9 is fixed by equating the volume integral of the density distribution function to the mass number @xmath10 of the residual daughter nucleus . the distance s between any two nucleons belonging to the residual daughter nucleus and the emitted @xmath0 nucleus is given by @xmath11 while the interaction potential between any such two nucleons @xmath12 appearing in eqn.(1 ) is given by the m3y effective interaction . the total interaction energy @xmath13 between the @xmath0 nucleus and the residual daughter nucleus is equal to the sum of the nuclear interaction energy , the coulomb interaction energy and the centrifugal barrier . thus @xmath14 where @xmath15 is the reduced mass , @xmath16 is the mass number of the parent nucleus and m is the nucleon mass measured in the units of @xmath17 . assuming spherical charge distribution ( scd ) for the residual daughter nucleus and considering the @xmath0-paticle to be a point charge , the @xmath0-nucleus coulomb interaction potential @xmath18 is given by where @xmath20 and @xmath21 are the atomic numbers of the @xmath0-particle and the daughter nucleus respectively . the touching radial separation @xmath22 between the @xmath0-particle and the daughter nucleus is given by @xmath23 where @xmath24 and @xmath25 has been obtained using eqn.(4 ) . if point charge distribution ( pcd ) is also assumed for the residual daughter nucleus then the @xmath0-nucleus coulomb interaction potential has the simple form of @xmath26 for all r. the energetics allow spontaneous emission of @xmath0-particles only if the released energy @xmath27 is a positive quantity , where @xmath28 , @xmath29 and @xmath30 are the atomic masses of the parent nucleus , the emitted @xmath0-particle and the residual daughter nucleus , respectively , expressed in the units of energy . it is important to mention here that the correctness of predictions for possible decay modes , therefore , rests on the accuracy of the ground state masses of nuclei . in the present work , the half life of the parent nucleus against the split into an @xmath0 and a daughter is calculated using the wkb barrier penetration probability . the assault frequecy @xmath31 is obtained from the zero point vibration energy @xmath32 . the half life @xmath33 of the parent nucleus @xmath34 against its split into an @xmath0 @xmath35 and a daughter @xmath36 is given by @xmath37 [ 1 + \exp(k ) ] \label{seqn9}\ ] ] where the action integral @xmath38 within the wkb approximation is given by @xmath39}^{1/2 } dr \label{seqn10}\ ] ] where @xmath40 and @xmath41 are the two turning points of the wkb action integral determined from the equations @xmath42 for the present calculations , from a fit to a selected set of experimental data on @xmath0 emitters the following law @xcite was found for the zero point vibration energies for the @xmath0 decays @xmath43 which includes the shell and pairing effects . the values of the proportionality constants of @xmath44 with @xmath45 is the largest for even - even parent and the smallest for the odd - odd one . if all other conditions are the same one may observe that with greater value of @xmath44 , the life time is shortened indicating higher emission rate . the shell effects of @xmath0 radioactivity is implicitly contained in the zero point vibration energy due to its proportionality with the q value , which is maximum when the daughter nucleus has a magic number of neutrons and protons . the two turning points of the action integral given by eqn.(10 ) have been obtained by solving eqns.(11 ) using the microscopic double folding potential given by eqn.(1 ) along with the coulomb potential given by eqn.(7 ) and the centrifugal barrier . then the wkb action integral between these two turning points has been evaluated numerically using eqn.(1 ) , eqn.(6 ) , eqn.(7 ) , eqn.(8 ) and eqn.(12 ) . the calculations have been performed using @xmath12 , inside the integral of eqn.(1 ) for the double folding ( df ) procedure , as only the m3y effective @xcite interaction @xmath46 this interaction is based upon a realistic g - matrix . since the g - matrix was constructed in an oscillator representation , it is effectively an average over a range of nuclear densities and therefore the m3y has no explicit density dependence . for the same reason there is also an average over energy and the m3y has no explicit energy dependence either . the only energy dependent effects that arises from its use is a rather weak one contained in an approximate treatment of single - nucleon knock - on exchange . the success of the extensive analysis @xcite indicates that these two averages are adequate for the real part of the optical potential for heavy ions at energies per nucleon of @xmath47 . however , it is important to consider the density and energy dependence explicitly for the analysis of @xmath0-particle scattering at higher energies ( @xmath48 ) where the effects of a nuclear rainbow are seen and hence the scattering becomes sensitive to the potential at small radii . such cases were studied introducing suitable and semirealistic explicit density dependence @xcite into the m3y interaction which was then called the ddm3y and was very successful for interpreting consistently the high energy elastic @xmath0 scattering data . the entire calculations have been redone again using the density dependent m3y effective interaction ( ddm3y ) supplemented by a zero - range pseudo potential , representing the single nucleon exchange term @xcite . in ddm3y the effective nucleon - nucleon interaction @xmath12 is assumed to be density and energy dependent and therefore becomes functions of density and energy and is generally written as @xmath49 where @xmath50 is the same m3y interaction given by eqn.(13 ) but supplemented by a zero range pseudo - potential @xmath51 where the zero - range pseudo - potential representing the single - nucleon exchange term is given by @xmath52 and the density dependent part has been taken to be @xcite @xmath53 which takes care of the higher order exchange effects and the pauli blocking effects . the energy e appearing in the above equations is the energy measured in the centre of mass of the @xmath0 - daughter nucleus system and for the @xmath0 decay process it is equal to the released energy q. since the released energies involved in the @xmath0 decay processes are very small compared to the energies involved in high energy @xmath0 scattering , @xmath54 has been considered as a constant and independent of energy . the zero - range pseudo - potential @xmath55 is also practically independent of energy for the @xmath0 decay processes and can be taken as @xmath56 . it is important to mention here that use of the nuclear interaction potential for exotic cluster decays obtained by double folding an effective nucleon - nucleon interaction with their respective densities was suggested in reference @xcite . after the wkb action integral @xmath38 , given by eqn.(10 ) has been evaluated , the half lives of the @xmath0 decays have been calculated using eqn.(9 ) and eqn.(12 ) . for the present illustrative calculations , the same set of experimental data of reference @xcite along with the rest from reference @xcite for the @xmath0 decay half lives have been chosen for comparison with the present theoretical calculations for even - even parent nuclei for which the experimental ground state masses for the parent and daughter nuclei are available . this set was selected because there is no uncertainty in the determination of the released energy q ( given by eqn.(8 ) ) which is one of the crucial quantity for quantitative prediction of decay half lives and all the the parent and daughter nuclei have zero spins and positive parities . for the calculations using ddm3y with zero range exchange also , the value of the normalization c appearing in eqn.(17 ) has been fixed at 1.0 . the density dependent parameter @xmath54 , which is supposed to be dependent on energy , has been kept constant and independent of q. optimum fit to the data has been obtained using @xmath54=1.6 . in fig . [ fig1 ] the experimental data for logarithmic @xmath0 decay half lives @xcite have been plotted against the mass numbers of parent nuclei along with the results of the present calculations for zero angular momentum of the fragments . in the figure the open circles depict the experimental data while the continuous line with solid circles represents the corresponding calculations of the present model using the m3y effective interaction and the dotted line represents the present calculations using ddm3y effective interaction supplemented by a zero - range pseudo - potential . the decay modes and the experimental values for their half lives have been presented in table 1 . the corresponding results of the present calculations of superasymmetric fission model with microscopic potentials are also presented along with the results of the liquid drop model ( ldm ) calculations of 2001 so as to facilitate the comparison of the results of the phenomenological calculations with the present one . the exact q values calculated by eqn.(8 ) using the experimental ground state masses @xcite and used by the present calculations have also been presented . the results of the present calculations with m3y or ddm3y with pseudopotential have been found to predict the general trend of experimental data very well . the quantitative agreement with experimental data is excellent . the degree of reliability of the present estimates for the @xmath0 decay lifetimes are better than the liquid drop description @xcite for most of the cases . the recent result of the preformed cluster model @xcite calculation for the @xmath0 decay half - life of @xmath57 is also much worse compared to that obtained from present calculations . the degree of reliability of the present estimates for the @xmath0 decay lifetimes is better than the recent results of the ldm calculations @xcite presented in the table below . .comparison between calculated @xmath0-decay half - lives using point ( pcd ) and spherical charge distributions ( scd ) respectively for the coulomb interaction and using effective interactions of m3y and ddm3y with zero - range pseudo - potential for the nuclear interaction . [ cols="^,^,^,^,^,^,^ " , ] @xmath58 taken from reference @xcite . above results show that the differences in the @xmath0-decay lifetimes obtained using the spherical charge distributions or point charge distributions for calculating the coulomb interaction are small . introducing the density - dependence and the zero - range pseudopotential do not alter the general trend of the results and also do not improve the quality of fit to the experimental data significantly . use of the m3y effective interaction alone for calculating the double folding nuclear interaction potentials is sufficient to provide reasonable estimates for the @xmath0-decay lifetimes . the half lives for @xmath0-radioactivity have been analyzed with microscopic nuclear potentials obtained by the double folding pocedure using m3y and ddm3y effective interactions respectively . this procedure of obtaining nuclear interaction potentials is based on profound theoretical basis . the results of the present calculations using m3y or ddm3y supplemented by a pseudo - potential are in good agreement over a wide range of experimental data . present calculations show that the differences in the results of the @xmath0 decay lifetimes obtained using the spherical charge distributions or point charge distributions for calculating the coulomb interaction are small . use of the density - dependence and the zero - range pseudopotential also neither alters the general trend of the results nor improves significantly the quality of fit to the experimental data . refinements such as introduction of dissipation while tunneling through the barrier or incorporating the dynamic shape deformations in the density distributions of the clusters may further improve results . it is worthwhile to mention that using the realistic microscopic nuclear interaction potentials , the results obtained for the @xmath0 radioactive decay lifetimes are noteworthy and are comparable to the best available theoretical calculations . such calculations may be extended to provide reasonable estimates of the lifetimes of nuclear decays by @xmath0 emissions for the entire domain of exotic nuclei .	radioactive decay of nuclei via emission of @xmath0 particles has been studied theoretically in the framework of a superasymmetric fission model using the double folding ( df ) procedure for obtaining the @xmath0-nucleus interaction potential . the df nuclear potential has been obtained by folding in the density distribution functions of the @xmath0 nucleus and the daughter nucleus with a realistic effective interaction . the m3y effective interaction has been used for calculating the nuclear interaction potential which has been supplemented by a zero - range pseudo - potential for exchange along with the density dependence . the nuclear microscopic @xmath0-nucleus potential thus obtained has been used along with the coulomb interaction potential to calculate the action integral within the wkb approximation . this subsequently yields calculations for the half lives of @xmath0 decays of nuclei . the density dependence and the exchange effects have not been found to be very significant . these calculations provide reasonable estimates for the lifetimes of @xmath0 radioactivity of nuclei .
it has been known that , for a given soliton equation , there exists a discrete analogue of the evolution equation , which preserves integrability . once we find a discrete integrable equation , we can derive infinitely many integrable differential equations by taking continuous limits of variables in many different ways . we are interested in characterizing such integrable discrete systems . in our previous papers @xcite we studied various integrable discrete systems which have certain symmetry and satisfy periodic boundary conditions . in this approach the nature of a quadratic equation , imposed by the boundary condition , plays a crucial role . the map which generates a time evolution turns out to be twofold , corresponding to two solutions of the quadratic equation . we showed that the discriminant of the quadratic equation becomes a perfect square when the system is integrable . owing to this fact the map is free from a square root , hence is rational . we will call this type of map a non square root map ( nsrm ) in what follows . the discrete versions of lotka - volterra equation , kdv equation , toda lattice , kp equation and painlev equations @xcite are such examples . the purpose of this paper is to generalize this scheme of characterizing discrete integrable systems . to clarify the point of our argument let us show first a typical example . the 3-point discrete time toda lattice is given by the set of coupled equations : @xmath1 here @xmath2 means the variable @xmath3 at the time @xmath4 , _ i.e. _ , @xmath5 . we impose the periodic conditions @xmath6 and @xmath7 . hence there are 6 dependent variables . before solving eq.([eqn : dtoda ] ) directly it is more instructive to solve @xmath8 for arbitrary functions @xmath9 of @xmath10 . we can also write them as @xmath11 note that they are special cases of mbius transformation . hence by repeating this transformation , say for the first equation of eq.([eqn : dtoda3 ] ) , three times , @xmath2 is given in terms of @xmath12 , which are nothing but @xmath13 themselves owing to the periodic boundary conditions . in this way we obtain an equation @xmath14 or , equivalently , @xmath15 where @xmath16 in the case of three point toda lattice . the quadratic equation eq.([quadratic eq ] ) has two solutions @xmath17 unless its discriminant @xmath18 vanishes . the general form of solutions contain square root of polynomial functions . under this circumstance a sequence of the map will yield very complicated orbits in general . but the integrable systems are not such cases , _ i.e. _ , the discriminant happens to be a perfect square of some polynomial . in fact if we substitute the right hand sides of eq.([eqn : dtoda ] ) into @xmath19 s and @xmath20 s of the expression eq.([abcd ] ) , the discriminant is considerably simplified and is given by @xmath21 this is a typical non square root map ( nsrm ) , which we are going to study ( see below for more precise definition ) . this remarkable feature could be understood from other point of view as well . namely by an inspection of our original equations eq.([eqn : dtoda ] ) we can see easily that @xmath22 is one of sets of solution . this implies that another set of solutions of the quadratic equation eq.([quadratic eq ] ) must not have a square root either . we would like to emphasize that the existence of the solution of such simple form eq.([x = u ] ) owes to the symmetry of the equation eq.([eqn : dtoda ] ) . we will study , in the following section , @xmath23-point maps under which the square roots are removed when periodic boundary conditions are imposed . in section 3 the discrete time toda equation is studied in detail based on the preceding argument . in particular we will show that this scheme enables one to generalize the toda equation to a @xmath0-difference form , so that various @xmath0-difference soliton equations are derived by reductions , including the @xmath24 equation by kajiwara , noumi and yamada@xcite . in section 4 , we will discuss time evolution of solutions . the last section is devoted to summarize arguments in this paper . we consider @xmath23-point systems in this section and study conditions under which a map becomes a non square root map ( nsrm ) . generalizing eq.([eqn : dtoda3 ] ) , let us consider the following sequence of mbius transformations , @xmath25 and mbius matrices @xmath26 after @xmath23 times of the transformation we obtain the equation @xmath27 while the corresponding mbius map is given by @xmath28 we now impose the periodic boundary conditions @xmath29 . then the formula eq.([n step map ] ) is equal to the quadratic equation @xmath30 the solutions of this equation are given by @xmath31 whereas the discriminant @xmath32 is @xmath33 in this subsection we study maps such that the discriminant eq.([dis ] ) becomes a perfect square of some polynomial function . as we will prove shortly the following 6 patterns of the map @xmath34 satisfy the requirement : @xmath35 @xmath36 proof : since the trace and the determinant are invariant under the transposition ot the matrices , we do not consider here the fourth and sixth patterns . for the first three patterns we find @xmath37 where @xmath38 and @xmath39 is a certain function of @xmath40 . from this expression we easily find @xmath41 corresponding to the first , second and the third patterns . for the 5th pattern we first observe @xmath42 in order to calculate @xmath43 , we split @xmath34 into two parts @xmath44 using the property @xmath45 we can see that @xmath46 has terms of the form @xmath47 only . from this fact , the trace of @xmath46 is calculated as @xmath48 where @xmath49 . similarly we can treat the 6th pattern and obtain the same results as eqs.([det of 5th ] ) and ( [ trace of 5th ] ) for the determinant and trace , respectively . hence we have obtained discriminants for all six patterns as follows : [ cols="<,<,<,^ , < " , ] therefore every discrete equation with the mbius matrix pattern of eq.([nonsrs - pat ] ) has no square root , although they are not all . solutions of the quadratic equation eq.([quadra eq ] ) are given by eq.([sol : quad ] ) . the purpose of this subsection is to write them explicitly in the case of nsrm . first we note that the 2nd and the 3rd pattern as well as even @xmath23 case of the 1st pattern do not form a quadratic equation due to @xmath50 . hence it is enough to consider the odd @xmath23 case of 1st pattern , 4th , 5th and 6th patterns . the odd @xmath23 case is given by the following setup : @xmath51 hence the solutions of the quadratic equation are @xmath52 in this pattern we have @xmath53 to calculate @xmath39 , we split @xmath34 to two parts @xmath54 and @xmath55 as follows : @xmath56 using the fact @xmath57 , we see that @xmath46 has terms of the form @xmath58 only , where @xmath59 and @xmath60 . all possible terms can be written in such forms as @xmath61 hence we find the expression of @xmath39 as @xmath62 the solutions of the quadratic equation are then given as follows : @xmath63 we already got @xmath43 and @xmath64 in the previous subsection , but did not have calculated the elements of @xmath46 . to obtain the solution of the quadratic equation , we must find their elements since information about @xmath65 and @xmath66 are necessary for the solutions . the splitting idea of eq.([split:5th ] ) is convenient to obtain the elements here again . using this idea , we need to calculate only the following quantities @xmath67 because @xmath68 . hence @xmath69 and @xmath70 of the elements of @xmath46 are obtained as @xmath71 consequently , noting the fact @xmath72 , we obtain the solutions of the quadratic equation as @xmath73 where @xmath74 is the homogeneous polynomial of @xmath75 , that is @xmath76 for this pattern too , we already got @xmath43 and @xmath64 in the previous subsection , but did not have calculated the elements of @xmath46 . using the splitting idea again , the elements of @xmath46 in this pattern are able to be obtained . the splitting of @xmath34 in this case is @xmath77 since @xmath78 , we are enough to calculate the following terms @xmath79 hence we obtain @xmath69 and @xmath70 of the elements of @xmath46 as @xmath80 if we further use @xmath81 , we get solutions of the quadratic equation as follows : @xmath82 we note that polynomial @xmath83 s , which appear in the solutions of the 5th and 6th patterns , are given by the minor determinants of the diagonal entries of the following matrix . @xmath84 for simplicity , we will introduce the following notations . @xmath85 namely , the matrix @xmath83 having entries in ascending order is expressed by @xmath83 , and in descending order by @xmath86 . these matrices are related by @xmath87 where @xmath88 we set symbols of them of @xmath89-entry as @xmath90 , _ i.e. _ @xmath91 where @xmath92 . these polynomials satisfy the following useful relations . 1 . @xmath93,[q = q^v ] 2 . @xmath94[(b+c)q = bq+cq ] . the relation @xmath95 is obvious . the relation @xmath96 is non - trivial and will be checked by means of more fundamental relation : @xmath97 where @xmath98 . we can show that @xmath99 coincides with the cofactor of the @xmath89 element of the product @xmath100 of the following matrices @xmath101 which could represent the dtoda equations @xcite . the discriminant @xmath32 is invariant under arbitrary similarity transformations , since it is given by the trace and determinant of @xmath46 . this fact enables us to modify our map starting from the six patterns we considered above . namely under the transformation of @xmath34 , @xmath102 the matrix eq.([s = mm ... m ] ) is changed into @xmath103 if we take into account the periodic boundary condition , we must have @xmath104 . therefore the trace and the determinant of @xmath46 , hence the discriminant @xmath32 as well , remain constant : @xmath105 under the transformation eq.([transf : mod ] ) for arbitrary matrices @xmath106 s . this implies that the modified map @xmath107 is nsrm if @xmath34 is so . now denoting by @xmath108 the modified variable of @xmath109 , we will write the mbius map associated with @xmath34 as @xmath110 . in this way , the modified mbius map is possible to be written as @xmath111 here if we define @xmath112 , then @xmath113 as a result , the following two modifications are equivalent @xmath114 this correspondence is preserved in the solutions of the quadratic equation , so we can create many equations from the basic six patterns in this way . we study , in this section , the @xmath23-point discrete time toda equation ( dtoda ) in detail from the view point of the previous section . let us start with writing the @xmath23-point dtoda equation . @xmath115 in order to apply the argument in the previous section to this system we transform eq.([eqn : n - point dtoda ] ) into the form of mbius transformation , @xmath116 corresponding to @xmath117 . we notice that this is a modified 5th pattern mbius transformation which is specified by the following data : @xmath118 substituting these data into eq.([sol:5th ] ) we obtain solutions immediately . before writing them down , however , it is convenient to simplify notations . namely the matrix @xmath83 , which appears in the solutions , is @xmath119 . let the matrix @xmath83 be @xmath120 and its minor determinants be @xmath121 these minor determinants satisfy 1 . @xmath122 , 2 . @xmath123 3 . @xmath124 where @xmath125 . therefore the solutions turn out to be @xmath126 similarly the time reversal map is obtained by writing eq.([eqn : n - point dtoda ] ) as @xmath127 corresponding to the map @xmath128 . this is again a modified 5th pattern of the mbius transformation . the modification matrix and the elements of the map are given by @xmath129 and the solutions @xmath130 instead of eq.([mm : toda 1 ] ) and eq.([mm : toda 1 inverse ] ) we could write the dtoda equation eq.([eqn : n - point dtoda ] ) in the form @xmath131 corresponding to @xmath132 and @xmath133 . these are modified 6th patterns whose elements and the modification matrix are given by @xmath134 for eq.([mm : toda 2 ] ) and eq.([mm : toda 2 inverse ] ) , respectively . applying these data to eq.([sol:6th ] ) we obtain the same results of eq.([b - type : toda ] ) and eq.([b - type : toda inv ] ) . we have seen that the b - type solution is simply 1 if the map @xmath34 is the form of either 5th or 6th pattern . a modification of the matrix by @xmath106 will change the solutions as well as the equation while the discriminant remains constant . in particular a different b - type solution will correspond to a different set of equations , and hence different a - type solution . if the b - type solution is rational the a - type solution must be also rational satisfying the same equations . to see how it works we try to generalize the b - type solution eq.([b - type : toda ] ) to @xmath135 it is not difficult to find their corresponding equations @xmath136 from the construction this equation has also a - type solution which is a rational polynomial of @xmath137 and @xmath138 . and we can set @xmath139 without loss of generality . we would like to emphasize that the b - type solution follows automatically to the symmetry possessed by the equation . in other words , the rational solution of a - type is a consequence of the symmetry of the toda lattice equation . in the paper @xcite kajiwara _ et al . _ have studied affine - weyl symmetry of the discrete time painlev iv equation with parameters , which is called @xmath0p@xmath140 . we have shown in our previous paper @xcite that @xmath0p@xmath140 is one of the equations which are characterized by the square root free maps . now we want to know whether exist similar generalization of the discrete time toda lattice eq.([eqn : n - point dtoda ] ) which we have been studying . to this end we consider the following equation with a set of parameters @xmath141 : @xmath142 let us call this equation a @xmath0-difference generalized discrete time toda equation ( @xmath0toda ) . remarkably this equation has the symmetry of modified 5th or 6th pattern of the mbius map so , this is transformed to a quadratic equation of every @xmath143 or @xmath138 , and has two types of map as solutions to eq.([qtoda ] ) . in fact if we introduce @xmath144 we obtain two maps as follows : @xmath145 where @xmath146 , and @xmath147 is defined by @xmath148 we can further generalize @xmath0toda to higher dimensions by increasing the number of suffices : @xmath149 the b - type map should have the form @xmath150 in the case of @xmath151 , for instance , @xmath152 satisfies @xmath153 this is the @xmath0kp equation which we derived in our previous paper . it has been well known that the discrete time lotka - volterra equation ( dlv ) can be obtained from the discrete time toda equation by a miura transformation @xcite . we like to show , in this section , that the same dlv - type as well as the discrete time kdv ( dkdv)-type equations can be derived from the discrete time toda lattice eq.([qtoda ] ) by simple reductions . the basic idea of our reduction is to deal with one of two equations in @xmath0toda as a constraint to eliminate half of the variables . the way of reduction to a dlv - type equation is to impose an additional constraint to the second equation of @xmath0toda as follows . @xmath154 where @xmath155 s are constants . this constraint enables us to eliminate variables , say @xmath138 and @xmath156 , from the first equation . namely the following substitution @xmath157 yields the dlv - type equation @xmath158 when @xmath159 , this is the dlv equation . various types of @xmath0lv equations which were discussed in our previous paper , are included in this equation . it is straightforward to obtain two maps of this equation . we find @xmath160 which could be also obtained from the map of qtoda eq.([qtoda ] ) by the substitution eq.([substitution ] ) . the @xmath0p@xmath161 of kajiwara , noumi , yamada @xcite is included as one of the a - type map . let us try an alternative case of constraint . namely we impose a constraint to the first equation of eq.([qtoda ] ) . @xmath162 this enables us to eliminate @xmath138 and @xmath156 from the second equation according to the rule @xmath163 from which we obtain @xmath164 this equation containes the discrete time kdv equation as a special case . hence we call eq.([qdkdv ] ) the @xmath0kdv equation . the corresponding solutions are given again by simply replacing @xmath138 and @xmath156 in the solutions of @xmath0toda according to eq.([qkdv reduction rule ] ) . @xmath165 we define some symbols of the two types of map and their inverse maps of dtoda as follows , @xmath166 these maps satisfy the following relations . @xmath167 or @xmath168 where @xmath169 and @xmath170 . the relations eq.([cd ] ) are derived from eq.([ab ] ) by some calculations , automatically . the reflection maps @xmath171 change the inverse a - type map to the inverse b - type one and the a - type one to the b - type one , respectively . namely , @xmath172 hence , these reflection maps are permutations of solutions of the quadratic equation eq.([quadra eq ] ) in the case of dtoda , _ i.e. _ galois tranformations of eq.([quadra eq ] ) . furthermore , the generalized version of dtoda eq.([gdtoda ] ) satisfy these relations eq.([ab ] ) or eq.([cd ] ) , but @xmath0toda does not . these reflection maps play an important role in affine - wyel symmetries and @xmath0-difference painlev equations as studied by kajiwara , _ et al_. @xcite . a sequence of b - type map does not generate an orbit but causes exchange of the variables @xmath137 and @xmath138 , whereas an a - type map generates an orbit . we will study a discrete time evolution of a - type map in this section . to make the time dependence explicit we write the variables @xmath173 and @xmath156 as @xmath174 and @xmath175 in the following . a repeated use of eq.([b - type : toda ] ) enables one to express @xmath176 and @xmath177 in terms of @xmath178 s and @xmath179 s as @xmath180 here we denote by @xmath181 the following polynomial of @xmath182 s and @xmath183 s , @xmath184 if we substitute eq.([x^t+1 , u^t+1 ] ) into the right hand side of eq.([q_k^t ] ) we will find a considerable cancellation of factors in the numerator and the denominator . the cancellation might be associated with a reduction of algebraic entropy @xcite . as a consequence we obtain a recurrence equation for @xmath185 : @xmath186 the a - type solutions are given if @xmath185 is found by solving eq.([iteration ] ) iteratively . the generalization of this result to the case of @xmath0toda is straightforward . this is , however , a cumbersome task . and so , we will look orbits of a - type map described by numerical computation . it is not difficult to see behaviour of orbits of the a - type map eq.([x^t+1 , u^t+1 ] ) as long as we generate them numerically by using computers . they are constrained on curves since there are sufficient number of conserved quantities . if parameters @xmath187 s and @xmath188 s are introduced the orbits tend to expand and closed curves turn to spirals . these are behaviours which we could expect naively from our previous experiences . instead of studying them in detail we would like to present here behaviour of ultra - discrete version of solutions . it has been known that if a discrete integrable equation has its ultra - discrete version the integrability is preserved @xcite . there exists a systematic method to ultra - discretize a discrete integrable equation if it contains only product and summation of variables but not subtractions . namely we introduce a new variable @xmath189 and replace @xmath137 by @xmath190 . the ultra - discretization procedure @xcite is defined by @xmath191 to be specific let us consider the case of dkdv - type equation . the @xmath0-difference version has been given by eq.([qdkdv ] ) . if we further specify to 3-point case the equations are @xmath192 @xmath193 the corresponding a - type map can be also found readily from the one of @xmath0-toda equation by the reduction eq.([qkdv reduction rule ] ) . @xmath194 @xmath193 the ultra - discrete version of eq.([3-qkdv - a ] ) turns to @xmath195 [ fig : udkdv00 ] [ fig : udkdv01 ] [ fig : udkdv02 ] [ fig : udkdv03 ] here we set @xmath196 . let @xmath197 s and @xmath198 s be rational numbers and impose a condition @xmath199 so that the map returns to initial values after some steps . under these circumstances we find maps such as presented in figures [ fig : udkdv00 ] to [ fig : udkdv03 ] . figure 1 shows the case of @xmath200 for all j=0,1,2 . similarly fig . 2 corresponds to the case of @xmath199 and @xmath201 . the straight lines in these figures are known to correspond to integrable orbits @xcite . figures 3 and 4 are the cases of @xmath202 and @xmath203 respectively . they do not seem to describe integrable orbits . therefore the orbits behave differently for different value of @xmath204 . this must be contrasted with the case of dkdv in which orbits form smooth closed curves under the condition @xmath205 for all @xmath206 . it will be worth while to point out that the dkdv map eq.([3-qkdv - a ] ) in the case of @xmath201 is invariant under the following bcklund transformation @xmath207 where @xmath208 @xmath209 and @xmath210 and @xmath211 . this is the same affine weyl symmetry shared by @xmath0p@xmath140 @xcite . but @xmath23(@xmath2124)-point dkdv maps of @xmath201 does not have this symmetry even though it exhibits an integrable orbit . we have also studied 3-point ultra - discrete lv - type maps in the same manner . the results are similar to those of kdv - type maps and found one integrable map corresponding to the @xmath0p@xmath140 . the main purpose of this paper was to clarify the method characterizing discrete integrable systems , which we introduced in our porevious works . we have shown that , in the case of integrable systems , two adjacent variables are related by a mbius transformation of special forms . we found at least 6 patterns of such forms . this observation enabled us to obtain quite large class of new discrete integrable systems . the generalized discrete toda equations ( dtoda ) of eq.([gdtoda ] ) is one of such examples . by certain reductions of the dtoda equation dkdv and dlv were obtained . the method admits introduction of arbitrary parameters into the equations and further generalization to @xmath0toda systems . if we solve the equations in the form in which each variable is given as a function of variables in the previous time , there exist always two types of solutions which we called a - type and b - type . the b - type map does not create orbit but exchanges variables , while a sequence of the a - type map generates an orbit . the map of a - type is written as a rational polynomial of variables whose numerator and denominator are expressed by minor determinants of the matrix @xmath83 in eq.([q ] ) . it is interesting to notice that this matrix @xmath83 is the same one known in the field of affine weyl group and @xmath0-painlev equations @xcite . we have not succeeded analyzing full behaviour of a sequence of a - type maps . it is one of the most interesting problems to find behaviour of the map when parameters @xmath187 s and @xmath188 s are introduced . some numerical results using ultra - discrete method have shown that the introduction of the parameters preserve integrability in some cases but not always . this result seems to show that the non square root map condition is not sufficient but necessary for a discrete map being integrable . it is desirable to clarify this point further . r. hirota , s. tsujimoto and t. imai : _ difference scheme of soliton equations _ " , in _ future directions of nonlinear dynamics in physical and biological systems _ , ed . by p.l.christiansen at al . , ( plenum press , new york , 1993 ) p.7 . y. ohta , r. hirota and s. tsujimoto : j.phys . * 62 * ( 1993 ) 1872 . r. hirota , and s. tsujimoto : j.phys.soc.jpn . * 64 * ( 1995 ) 3125 - 3127 m. jimbo and t. miwa : j. phys . soc . jpn . * 51 * ( 1982 ) 4116 , 4125 . b. kupershmidt : _ kp or mkp _ mathematical surveys and monographs vol * 78 * ( american mathematical society , 2000 ) and references therein . y. suris : see , for instance , arxiv : solv - int/9902003 .	we have proposed , in our previous papers@xcite , a method to characterize integrable discrete soliton equations . in this paper we generalize the method further and obtain a @xmath0-difference toda equation , from which we can derive various @xmath0-difference soliton equations by reductions .
general relativity ( gr ) is a self - consistent theory of gravity where space and time are considered as dynamical variables and new concepts as black holes and cosmic expansion are introduced . from the cosmological point of view , the prediction of cosmic microwave background radiation ( cmbr ) and the formation of primordial light elements maybe represent the greatest success of this theory . however , despite all these fundamental results , gr has not been fully investigated at the ultraviolet scales , where strong deviations from the standard hilbert - einstein picture emerge , and , on the other hand , new ingredients , such as dark matter ( dm ) and dark energy ( de ) , are required in order to fit the gravitational dynamics at infrared scales @xcite . moreover , self - consistent and comprehensive approaches to deal with gravitational interactions at fundamental level ( quantum gravity ) are still missing . these arguments lead to the conclusion that a unitary theory encompassing the gravitational phenomenology at all scales is still lacking . in the last years , several _ alternative _ or _ modified _ theories of gravity have been proposed , also with the aim to address the shortcomings related to the cosmological standard model , based on gr . for example , higher order curvature invariants than the simple ricci scalar @xmath2 allow to get inflationary behaviors , removing the primordial singularity , as well as to explain the flatness and horizon problems @xcite . this approach and , of course , all those related to it , are fundamentally motivated by the fact that , at high curvature regimes , further curvature invariants have to be considered for constructing self - consistent effective actions in curved spacetime @xcite . in some sense , the introduction of higher order terms , depending on the invariants of curvature , are required at high curvature regimes . of course , this is not the final step for building up a quantum gravity theory , but it allows for an effective description that works well at least at one - loop level @xcite . all the above motivations strongly suggest that towards ultraviolet regimes ( i.e. in high density regimes ) , gr has to be modified by adding further curvature corrections . in the framework of models that extend gr , @xmath0 gravity is certainly one of the favorite candidate since it provides , in a natural way , an almost unified description of de and dm , without invoking exotic sources as dm @xcite . moreover , it allows for the unification of the early - time ( inflation ) and the later - time acceleration of the universe @xcite . the gravitational action for @xmath0 gravity is given by @xmath3\,,\ ] ] where @xmath4 is the action of the standard matter and @xmath5 , with the planck mass @xmath6gev ( details and applications are discussed in @xcite ) . this theory can be viewed as a particular case of scalar - tensor gravity by dealing with further degrees of freedom as a scalar field ( see @xcite for details ) . one of the consequences of dealing with alternative cosmologies , including hence @xmath0 cosmology , is that the thermal history of particles results modified . in fact , one finds that the expansion rates @xmath7 of the universe , obtained in modified cosmologies , can be written in terms of the expansion rate @xmath8 obtained in gr , @xmath9 , where the factor @xmath10 encodes the information about the particular model of gravity extending or modifying gr . usually , the factor @xmath10 is defined in order that the successful predictions of the big bang nucleosynthesis ( bbn ) are preserved , that is @xmath11 at early time , and @xmath12 before bbn begins ( one refers to the pre - bbn epoch since it is not directly constrained by cosmological observations ) . the aim of this paper is to explore the possibility of explaining the pamela observations , i.e. the measured excess in positron flux above @xmath13gev @xcite , in terms of dm thermal relic abundance , by means of @xmath0 cosmology . in fact , since in the framework of the conventional cosmology and particle physics a widely accepted explanation for such observations is still lacking , one , in principle , can not rule out dark matter , as weakly interactive massive particles ( wimps ) , axions or heavy neutrinos , as a possible solution . it is worth noting that dm interpretation of pamela data has indeed renewed the interest in alternative cosmologies because , as already noted , these models lead in a natural way to a modification of the expansion rate of the universe @xcite : an enhanced pre - bbn expansion can reconcile observed dm cosmic relic abundance with indirect dm detection experiments ( such as pamela and the more recent ams-02 ) . according to the above considerations , we assume that the evolution of the universe is governed by an @xmath0 model of the form @xmath14 this model can be generated , for example , in the framework of supergravity @xcite , and in a perturbative regime it turns out to be a correction of the @xmath15 model ( extended starobinsky s model ) . results in refs . @xcite show that the starobinsky model is in good agreement with the bicep2 data . moreover , sizable primordial tensor modes can be generated in ( marginally deformed ) models of the form ( [ f(r)=r+arn ] ) , provided @xmath16 @xcite ( more specifically in @xcite it is shown that if inflation is driven by @xmath0 gravity , then a natural form for this function is ( [ f(r)=r+arn ] ) , where the value of @xmath17 could be related to the microscopic theory dictating the trace - log quantum corrections ) . the main point is to rewrite the @xmath0 action in the einstein frame , which implies the appearance of a scalar field . the derivation relies essentially on two steps : @xmath18 the introduction of the conformal mode @xmath19 and of the real scalar field @xmath20 , related to @xmath21 as @xmath22 ; @xmath23 the generation of the kinetic term for @xmath20 through the conformal transformation @xmath24 . inflation occurs for large values of the scalar field , i.e. @xmath25 . this implies @xmath26 or , equivalently , @xmath27 , therefore a regime where the @xmath28-term is dominant . moreover , the starobinsky - like inflation model may also emerge from dilaton dynamics in brane cosmology scenarios based on string theory @xcite . models based on ( [ f(r)=r+arn ] ) have been also studied in the context of bouncing cosmology ( see , for example , @xcite ) . it has to be also mentioned that the recent analysis by the planck collaboration @xcite led to the conclusion that @xmath29-inflation ( @xmath30 ) is fully consistent with observations @xcite . at this point is worth a comment about the _ chameleon mechanism _ and @xmath0 gravity @xcite . such a mechanism asserts that the compton wavelength @xmath31 ( typically assumed constant ) , associated to the characteristic scales obtained by adding ( pertubative ) higher - order terms to the hilbert - einstein action , is smaller / larger in those regions where the matter density is higher / lower . as a consequence , the theory can be seen as a local effective theory which is valid for a certain range of parameters . for @xmath0 gravity , the chameleon mechanism works because , in the high - energy density regions , it reproduces the newtonian gravitational forces , making the model compatible with the solar system tests @xcite ( see also @xcite ) . the paper is organized as follows . in section ii we derive the @xmath0 gravity field equations and solve them in the radiation dominated era ( well before the bbn onsets ) , hence from the period of gut scales to the transition time @xmath32 , when the universe starts to evolve according to the standard cosmological model . section iii is devoted to the study of thermal relics abundance and dm particles required to explain the pamela experiment . conclusions are drawn in section iv . the field equations for @xmath0 gravity follow by varying the action ( [ lagr ] ) with respect to the tensor metric @xmath33 @xmath34 where @xmath35 , and @xmath36 is the energy - momentum tensor for matter . the equation of the trace is @xmath37 the tensor @xmath38 satisfies the bianchi identities , i.e. @xmath39 , so that , for consistency , one gets that @xmath40 is divergenceless too : @xmath41 in a ( spatially flat ) friedman - robertson - walker ( frw ) metric @xmath42\,,\ ] ] the nonvanishing components of @xmath43 are @xmath44 where we have used @xmath45 , @xmath46 , and the dot stands for @xmath47 . as specified in the introduction , we work in the regime where the @xmath28-term dominates ( @xmath48 in eq . ( [ f(r)=r+arn ] ) , with @xmath16 according to ref . @xcite ) during the universe evolution from gut scales to the transition time @xmath49 . the latter characterizes the instant in which the universe passes from the cosmic evolution described by @xmath0 cosmology to the cosmic evolution described by the standard cosmological model ( see below ) . for simplicity , we look for solutions of the form @xmath50 . the @xmath51 field equation and the trace equation give ( in the early universe , hence in the limit @xmath52 ) @xmath53 where @xmath54 is the energy density , that in the radiation dominated era reads @xmath55 ( @xmath56 counts the number of relativistic degrees of freedom and @xmath57 is the temperature ) , meanwhile @xmath58\ , , \label{omega}\\ \gamma_{\beta , n } & \equiv & n-2-\frac{n(n-1)(2n-1)}{\beta(2\beta-1)}+\frac{3n(n-1)}{2\beta-1}\ , , \label{gamma } \\ r & = & \frac{6\beta(1 - 2\beta)}{t^2}\ , . \label{rdef } \end{aligned}\ ] ] the functions @xmath59 vs @xmath60 are plotted in figs . [ figo ] and [ figg ] . from ( [ hmodified ] ) and ( [ hmodifiedtrace ] ) , it follows that their ratio is given by @xmath61 . in what follows we shall consider two cases , @xmath62 and @xmath63 ( in the first case one gets a relation between @xmath60 and @xmath17 , in the second case @xmath60 and @xmath17 can be taken independent ) : * @xmath62 , i.e. @xmath64 - in this case , one gets two solutions : @xmath65 for these solutions , the function @xmath66 assumes the form @xmath67 and @xmath68 . the function @xmath69 is positive for @xmath70 . * @xmath71 - this possibility may occur , for example , in scenarios where bulk viscosity effects are considered , with @xmath72 , can be obtained in the case in which , for example , the interactions among massless particles are taken into account @xcite . these give rise to a trace anomaly @xmath73 , so that the adiabatic index turns out to be @xmath74 , where @xmath75 at high energies and for typical gauge groups and matter content in the universe , one finds that the order of magnitude of this quantity is given by @xmath76 , playing therefore a not negligible role in the early phases of the universe evolution . another possibility to a have a non vanishing @xmath77 is to consider quantum fluctuations of primordial fields @xcite . for a frw universe one gets @xmath78 , where @xmath79 , where @xmath80 is the number of quantum fields ( for su(5 ) model one finds @xmath81 ) . however , this case can not be used in the present paper , since field equations are not fulfilled . ] . these effects are generated owing to the rapid expansion / compression of fluids , ceasing to be in thermodynamical equilibrium . this occurs , in particular , in an expanding universe , when fluids are out of equilibrium . typically , these processes are so rapid that the system undergoes thermal equilibrium very quickly . however , in the case in which one considers particle decays of one or more species ( see for example @xcite ) , then a finite time is required for driving the system at the equilibrium . the energy - momentum tensor in presence of bulk viscosity term is given by @xcite ( see also @xcite ) @xmath82 with trace @xmath83 . here @xmath84 is the four - velocity of the fluid ( @xmath85 ) , and @xmath86 the bulk viscous pressure . following @xcite , we concern here the case in which the bulk viscous pressure is proportional to @xmath54 , i.e. @xmath87 . hence , one gets @xmath88 . in fig . [ figgammapi ] is plotted @xmath89 vs @xmath60 for @xmath90 . vs @xmath60 for different values of @xmath91.,title="fig:",width=384 ] + vs @xmath60 for different values of @xmath91.,title="fig:",width=384 ] + vs @xmath60 for different values of @xmath90.,title="fig:",width=384 ] + let us now determine the relation between the cosmic time @xmath92 and the temperature @xmath57 . from eq . ( [ hmodified ] ) one obtains @xmath93 where because of reality of @xmath49 for all @xmath60 . although the field equations remain the same under the change @xmath94 , the quantity @xmath95 change sign in order that the @xmath28 is always well defined for all @xmath60 . ] @xmath96^{1/2}\left(\frac{15{\tilde \alpha}\omega_{\beta , n}}{4\pi^3 g_}\right)^{\frac{1}{2n}}\ , , \qquad { \tilde \alpha}=\frac{\alpha}{m_{pl}^{2(1-n)}}\,.\ ] ] the transition time ( temperature ) @xmath49 ( @xmath97 ) is determined by equating the equation of the evolution in @xmath0 cosmology , eq . ( [ hmodified ] ) , with that one in gr , i.e. @xmath98 one gets @xmath99^n\right]^{\frac{1}{2(n-1)}}m_{pl}^{-1}\,,\ ] ] the expression of the transition temperature @xmath97 is then given by @xmath100^{\frac{1}{4 } } \frac{\left[4{\tilde \alpha}\omega_{\beta , n}\right]^{-\frac{1}{4(n-1)}}}{[6(\|\beta(1 - 2\beta)\|)]^{\frac{n}{4(n-1 ) } } } m_{pl}\,,\ ] ] that allows to recast the relation ( [ t - transition ] ) in the form @xmath101 moreover , notice that @xmath102 at the end we recover that the expansion rate of the universe in @xmath0 cosmology can be written as @xmath103 where the factor @xmath10 is the so called enhancement factor . in this section we study the thermal relic abundance in @xmath0 gravity . as mentioned in the introduction , alternative cosmologies indeed predict modified thermal histories of relic particles that occur during the _ pre _ bbn epoch . when the expansion rate of the universe changes , as compared to that one derived in the framework of gr , thermal relics decouple with larger relic abundance , a scenario that might have its imprint on relic wimps ( weakly interacting massive particles ) . the interest about these particles as dm follows from the fact that wimps in chemical equilibrium have the same abundance of cold dm . these studies are also motivated by astrophysical results obtained for cosmic ray electron and positrons @xcite , antiprotons @xcite , and @xmath89-rays @xcite . as mentioned , particular attention is devoted to pamela experiment that observed a excess of positron fraction at energies greater than 10gev @xcite . besides the astrophysical interpretation of this phenomenon proposed in refs . @xcite , there is also the possibility that the raise of the positron fraction could be ascribed to dm annihilation into leptons @xcite . in this last case , a large value of @xmath104 is required . more specifically , pamela and atic data require a cross section of the order or larger than @xmath105 @xmath106 sec@xmath107 . such a value is also necessary in order that thermal relics have the observed dm density ( see also @xcite ) . the characteristics of the universe expansion , such as the composition and/or the expansion rate , affect the relic density of wimps ( and more generally of other dm candidates ) as well as their velocity distributions before structure formation . according to the standard cosmology and particle physics , the calculation of the relic density of particles relies on the assumption that the period of the universe dominated by radiation began before the main production of relics and that the entropy of matter is conserved during this epoch and the successive one . however , any contribution to the energy density ( in matter and _ geometrical _ sector ) modifies the hubble expansion rate , and , as a consequence , the value of the relic density . investigations along these lines have been performed in different cosmological scenarios @xcite . the general analysis that accounts for the enhancement of the expansion rates in alternative cosmology has been performed in ref . @xcite ( see also @xcite ) . the expansion rate @xmath7 is written in the form @xmath108 , where the function is parameterized as @xmath109 where @xmath110mev . this form of @xmath10 allows to avoid conflicts with bbn . in the regime @xmath111 , the function ( [ a(t ) ] ) behaviors as @xmath112 . ] @xmath113 encodes , through the free parameters @xmath114 , a particular cosmological model , @xmath115 is the temperature at which the wimps dm freezes - out in the standard cosmology , @xmath116gev ( @xmath115 , in general , varies by varying the dm mass @xmath117 ) . the parameter @xmath118 labels cosmological models : @xmath119 in randall - sundrum type ii brane cosmology @xcite , @xmath120 in the kination models @xcite , @xmath121 in cosmologies with an overall boost of the hubble expansion rate @xcite , @xmath122 in scalar - tensor cosmology @xcite . in our @xmath0 model , we have @xmath123 , so that @xmath124 for @xmath125 . for our estimations , we shall refer to the analysis performed in @xcite , where the conditions for which modified cosmologies can explain both the pamela data and the abundance of relics particles , without violating the constraints provided by astrophysical observations , have been studied . more specifically the analysis concerns the dm annihilation cross section @xmath126 vs dm mass in the interval [ 10gev-10tev ] ( these annihilation channels are typical of several dm particles , such as lightest susy or kaluza - klein particles ) , for different annihilation channels ( @xmath127 , @xmath128 , @xmath129 , @xmath130 , @xmath131 ) and for different dm density profiles ( via lactea and aquarius dm distributions ) . the study is performed by numerically solving the boltzmann equation for the number density of thermal relic , where @xmath132 is the critical density of the universe , @xmath133 is the present value of the entropy density , and @xmath134 is the present value of the wimp abundance for comoving volume @xcite @xmath135 here @xmath136 is the value of the wimp abundance for comoving volume at the freeze - out , @xmath137 counts the effective number of degrees of freedom at temperature @xmath57 , and @xmath138}$ ] , which is computed for non relativistic dm particles . notice that in @xmath0 cosmology one has @xmath139 , where @xmath140 . this factor does not alter the boltzmann equation since @xmath141 for @xmath142 and it is of the order @xmath143 , for the values @xmath144 and @xmath145 here used . we can therefore safety use results of ref . ] @xmath146 , taking into account for the modifications related to the expansion rate @xmath147 . fixing @xmath148 and @xmath149 @xmath106 sec@xmath107 , one determines the values of the parameter @xmath150 vs @xmath151 $ ] , required to infer the correct relic abundance of dm particles @xmath152 @xcite . the analysis in @xcite shows that the values of the parameter @xmath150 necessary to explain the pamela data ( in particular for the case of dm annihilation into @xmath153 ) , together with @xmath154 , are ( see figg . 11 , 12 , and 15 of @xcite ) @xmath155 more precisely , for dm masses in the range @xmath156$]gev , the allowed region for the parameter @xmath150 is @xmath157 , where the upper bounds on @xmath150 vary for the different cosmological models labelled by @xmath118 . according to the above results , we rewrite the factor @xmath10 ( see eq . ( [ atf ] ) ) in the following form @xmath158 the transition temperature @xmath97 is fixed for values greater than the free - out temperature @xmath115 . therefore we set @xmath159 . from ( [ ttransition ] ) we get @xmath160^{-n}}{4\omega_{\beta , n } } \left(\frac{m_{pl}}{t_}\right)^{4(n-1)}m_{pl}^{2(1-n)}\,.\ ] ] the order of magnitudes of @xmath161 are reported in table i for @xmath60 and @xmath66 given in ( [ omegabeta1 ] ) . .in this table are reported some estimations of @xmath161 for fixed values of the transition temperature @xmath162mev . the expressions of @xmath60 and @xmath66 are given in ( [ omegabeta1 ] ) . [ cols="^,^,^",options="header " , ] the function @xmath150 vs @xmath17 is plotted in fig . [ figetan ] for @xmath66 and @xmath142 given in ( [ omegabeta1 ] ) , corresponding to @xmath62 , and in fig . [ figeta ] for @xmath163 given in ( [ omega ] ) , corresponding to @xmath63 . in both cases , the parameter @xmath150 assumes values of the order @xmath164 , so that the mass of wimps particles is of the order @xmath165gev . notice finally that the enhancement factor ( [ a(t)new ] ) increases for larger values of @xmath60 , hence for a super - accelerated expansion of the early universe . vs @xmath17 for @xmath142 and transition temperatures @xmath166mev . @xmath167gev is the freeze - out temperature , while @xmath168 is the lower bound on @xmath150 , see eq . ( [ etavaluespamela]).,title="fig:",width=384 ] + with @xmath169.,title="fig:",width=384 ] + in this paper we have studied the problem of thermal relic particles in @xmath0 cosmology . we focus on a power law model of the form @xmath170 , which is consistent with recent planck collaboration and bicep2 data constraints : @xmath171 corresponds to starobinsky s model , while the so called marginally deformed model ( @xmath172 ) produces sizable primordial tensor modes provided the exponent @xmath17 falls down in the range @xmath173 $ ] . as we have showed , if the cosmic evolution of the early universe is described by modified field equations , as provided indeed by @xmath0 gravity , then the expansion rate gets modified by a factor @xmath10 ( @xmath174 , see eq . ( [ atf ] ) ) . this quantity essentially weights how much the expansion rate of the universe in @xmath0 cosmology deviates from the expansion rate derived in the standard cosmology , and affects , in turn , the production of relic particles ( thermal relics decouple with larger relic abundances ) . as a consequence , the latter is obtained for larger annihilation cross section , and therefore also the indirect detection rates get enhanced . this effect may have its imprint on supersymmetric candidates for dm . for a power law scale factor , solutions of the modified field equations , and parameterizing the enhancement factor as @xmath175 , we find that the @xmath0 model is consistent with pamela data ( for dm annihilation into lepton channel @xmath153 ) , and the abundance of relic dm @xmath154 , provided that @xmath176 ( and @xmath177 ) . according to ( [ etavaluespamela ] ) , the corresponding wimps masses are @xmath178gev . finally , it is worth noticing that the analysis here performed relies on the model in which the form of @xmath0 is a power - law expansion of the scalar curvature @xmath2 , as well as on power law solution of scale factor . in general , it would be interesting to consider other curvature invariants , such as the riemann , the ricci and the gauss - bonnet tensors , and their derivatives , which could play a relevant dynamical role for the evolution of relic particles , as well as to seek a more general solution of the field equations for the @xmath0 model here discussed . 99 s. capozziello and m. de laurentis phys.rept . * 509 * , 167 ( 2011 ) . a. a. starobinsky , phys . b * 91 * , 99 ( 1980 ) . n.d . birrell and p.c.w . davies , _ quantum fields in curved space _ , cambridge university press , cambridge , uk ( 1982 ) . buchbinder , s.d . odintsov , i.l . shapiro , _ effective action in quantum gravity _ , iop , bristol , uk ( 1992 ) . n.h . barth and s. christensen , phys . d * 28 * , 8 ( 1983 ) . s. capozziello and m. de laurentis , annalen phys . * 524 * , 545 ( 2012 ) . et al . _ , j. * 116 * , 1009 ( 1998 ) . s. perlmutter _ _ , nature ( london ) * 391 * , 51 ( 1998 ) . s. nojiri and s.d . odintsov , phys . rep . * 505 * , 59 ( 2011 ) . a. silvestri and m. trodden , rep . * 72 * , 096901 ( 2009 ) . frieman , m.s . turner , and d. huterer , annu . . astrophys . * 46 * , 385 ( 2008 ) . r. durrer and r. maartens , gen . . grav . * 40 * , 301 ( 2008 ) . s. capozziello and g. lambiase , e - print : arxiv:1304.5640 [ astro-ph.co ] . m. sami , lect.notes phys . * 720 * , 219 ( 2007 ) . copeland , m. sami , and sh . tsujikawa , int . j. mod * d * 15 , 1753 ( 2006 ) . l. amendola and s. tsujikawa , _ dark energy : theory and observations _ , cambridge university press , 2010 . s. capozziello and v. faraoni , _ beyond einstein gravity _ , springer , new york , ( 2011 ) . s. capozziello , int . j. mod d * 11 * , 483 ( 2002 ) . s. nojiri , s.d . odintsov , phys . d * 77 * , 026007 ( 2008 ) . h. oyaizu , m. lima , and w. hu , phys . d * 78 * , 123524 ( 2008 ) . l. pogosian and a. silvestri , phys . d * 77 * , 023503 ( 2008 ) . i. sawicki and w. hu , phys . d * 75 * , 127502 ( 2007 ) . b. li and j.d . barrow , phys . d * 75 * , 084010 ( 2007 ) . t. clifton , phys . d * 78 * , 083501 ( 2008 ) . t. clifton and j.d . barrow , phys . d * 72 * , 103005 ( 2005 ) . s. capozziello and g. lambiase , gen . 32 * , 295 ( 2000 ) . s. capozziello and g. lambiase , gen . . grav . * 31 * , 1005 ( 1999 ) . s. capozziello , g. lambiase , and h.j . schmidt , annalen phys . * 9 * , 39 ( 2000 ) . s. nojiri and s.d . odintsov , phys . d * 72 * , 023003 ( 2005 ) . s. capozziello , v.f . cardone , e. elizalde , s. nojiri , and s.d . odintsov , phys . d * 73 * , 043512 ( 2006 ) . i. brevik , e. elizalde , s. nojiri , and s.d . odintsov , phys . d * 84 * , 103508 ( 2011 ) . s. nojiri and s.d . odintsov , phys . b * 639 * , 144 ( 2006 ) . s. nojiri and s.d . odintsov , prog . * 190 * , 155 ( 2011 ) . sotiriou and v. faraoni , rev . phys . * 82 * , 451 ( 2010 ) . a. de felice and s. tsujikawa , living rev . rel . * 13 * , 3 ( 2010 ) . g. lambiase , phys . d * 90 * , 064050 ( 2014 ) . g. lambiase , s. mohanty , and a.r . prasanna , int . j. mod . d * 22 * , 1330030 ( 2013 ) . g. lambiase and g. scarpetta , phys . d * 74 * , 087504 ( 2006 ) . b. jain , v. vikram , and j. sakstein , astrophys . j. * 779 * , 39 ( 2013 ) .. l. lombriser , a. slosar , u. seljak , and w. hu , phys . d * 85 * , 124038 ( 2012 ) . s. derraro , f. schmidt , and w. hu , phys . d * 83 * , 063503 ( 2011 ) . l. lombriser , f. schmidt , t. baldauf , r. mandelbaum , u. seljak , and r.e . smith , phys . d * 85 * , 102001 ( 2012 ) . f. schmidt , a. vikhlinin , and w. hu , phys . d * 80 * , 083505 ( 2009 ) . h. motohashi , a.a . starobinsky , and j. yakoyama , phys . * 110 * , 121302 ( 2013 ) . o. adriani _ et al . _ , nature ( london ) * 458 * , 607 ( 2009 ) . r. catena , n. fornengo , m. pato , l. pieri , and a. masiero , phys . rev . d * 81 * , 123522 ( 2010 ) . s. ferrara , a. kehagias , and a. riotto , fortsch . * 62 * , 573 ( 2014 ) ; fortsch . phys . * 63 * , 2 ( 2015 ) . s. basilakos , j. a. sales lima , j. sola int . j. mod d * 23 * , 1442011 ( 2014 ) . a. codello , j. joergensen , f. sannino and o. svendsen , jhep * 02 * , 050 ( 2015 ) . j. ellis , n.e . mavromatos , d.v . nanopoulos , phys.lett . b * 732 * , 380 ( 2014 ) . k. bamba , a.n . makarenko , a.n . myagky , s. nojiri , and s. odintsov , jcap * 01 * , 008 ( 2014 ) . m. novello and s.e.p . bergliaffa , phys . rep . * 463 * , 127 ( 2008 ) . n. paul , n. chakrabartym and k. bhattacharya , jcap * 10 * , 009 ( 2014 ) . ( planck collaboration ) , a&a * 57 * , a22 ( 2014 ) . v.f . mukhanov and g. chibisov , jept lett . * 33 * , 532 ( 1981 ) . starobinsky , sov . * 9 * , 302 ( 1983 ) . j. khoury and a. weltman , phys . lett . * 93 * , 17104 ( 2004 ) ; phys . d * 69 * , 044026 ( 2004 ) . i. navarro and k. van acoleyen , j. cosmol . astropart . phys . 02 ( 2007 ) 022 . t. faulkner , m. tegmark , e.f . bunn , and y. mao , phys . d * 76 * , 063505 ( 2007 ) . t. tamaki and s. tsujikawa , phys . d * 78 * , 084028 ( 2008 ) . s. tsujikawa , t. tamaki , and r. tavakol , j. cosmol . phys . 05 ( 2009 ) 020 . w. hu and i. sawicki , phys . d * 76 * , 064004 ( 2007 ) . brax , c. van de bruck , a .- ch . davis , and d.j . shaw , phys . d * 78 * , 104021 ( 2008 ) . l. amendola , r. gannouji , d. polarski , and s. tsujikawa , phys . d * 75 * , 083504 ( 2007 ) . starobinsky , jetp lett . * 86 * , 157 ( 2007 ) . s. tsujikawa , phys . d * 77 * , 023507 ( 2008 ) . s. capozziello and s. tsujikawa , phys.rev . d * 77 * , 107501 ( 2008 ) . saaidi and a. aghnamohammadi , astrophys . space sci . * 333 * , 327 ( 2011 ) . n. kaloper , phys . b * 653 * , 109 ( 2007 ) . a. iglesias , n. kaloper , a. padilla , and m. park , phys . d * 76 * , 104001 ( 2007 ) . k. kajantie , m. laine , k. rummukainen and y. schroder , phys . d 67 , 105008 ( 2003 ) . h. davoudiasl , r. kitano , g.d . kribs , h. murayama , and p.j . steinhardt , phys . lett * 93 * , 201301 ( 2004 ) . r. opher and a. pelisan , phys d * 74 * , 023505 ( 2006 ) . g. lambiase and s. mohanty , phys . d * 84 * , 023509 ( 2011 ) . wilson , g.j . mathews , and g.m . fuller , phys . d * 75 * , 043521 ( 2007 ) . w. zimdahl , phys . d * 53 * , 5483 ( 1996 ) . w. zimdahl , phys . d * 57 * , 2245 ( 1998 ) . s. weinberg , _ gravitation and cosmology : principles and applications of the general theory of relativity _ , jhon wiley & sons , inc . , new york , 1972 ; astrophysical j. * 168 * , 175 ( 1971 ) . h. okumara and f. yonezawa , physica a * 321 * , 207 ( 2003 ) . hoorer , a.j.c . ladd , r.b . hickman , and b.l . holian , phys . a * 21 * , 1756 ( 1980 ) . hoover , d.j . evans , r.b . hickman , a.j.c . ladd , w.t . ashurst , and b. moran , phys . a * 22 * , 1690 ( 1980 ) . r. maartens , arxiv : astro - ph/9609119 . m. novello , j.b.s . dolival , acta physics polonica * 11 * , 3 ( 1980 ) . m. novello and r.a . araujo , phys . d * 22 * , 260 ( 1980 ) . grn , astrophysics and space science * 173 * , 191 ( 1990 ) . i. brevik , v.v . obukhov , and a.v . timoshkin , astrophys . space sci . * 355 * , 399 ( 2015 ) ; arxiv:1507.05216 . s. nojiri and s. odintsov , phys . b * 639 * , 144 ( 2006 ) . et al . _ , nature * 456 * , 362 ( 2008 ) . et al . _ [ the fermi lat collaboration ] , phys . 102 * , 181101 ( 2009 ) . f. aharonian _ _ [ h.e.s.s . collaboration ] , phys . * 101 * , 261104 ( 2008 ) ; astron . astrophys . * 508 * , 561 ( 2009 ) . o. adriani _ et al . _ , * 102 * , 051101 ( 2009 ) . f. aharonian _ et al . _ , * 97 * , 221102 ( 2006 ) . t.a . porter ( fermi lat collaboration ) , arxiv:0907.0294 [ astro - ph ] . a.a . abdo , astrophys . j. * 703 * , 1249 ( 2009 ) . d. hooper , p. blasi , p.d . serpico , jcap * 0901 * , 025 ( 2010 ) . p. blasi , phys . lett . * 103 * , 051104 ( 2009 ) . p. blasi , p.d . serpico , phys . * 103 * , 081103 ( 2009 ) . n. arkani - hamed , d.p . finkbeiner , t.r . slatyer , and n. weiner , phys . d * 79 * , 015014 ( 2009 ) . robertson and a.r . zentner , phys . d * 79 * , 083525 ( 2009 ) . m. cirelli , m. kadastik , m. raidal , and a. strumia , nucl . b * 813 * , 1 ( 2009 ) . i. cholis , l. goodenough , d. hooper , m. simet , and n. weiner , phys . d * 80 * , 123511 ( 2009 ) . f. donato , d. maurin , p. braun , t. delahaye , and p. salati , phys . lett . * 102 * , 071301 ( 2009 ) . k. ishiwata , s. matsumoto , and t. moroi , phys . b * 675 * , 446 ( 2009 ) ; phys . d * 78 * , 063505 ( 2008 ) . spergel , _ et al . _ , ( wmap collaboration ) , astrophys . j. suppl . * 170 * , 377 ( 2007 ) . m. schelke , r. catena , n. fornengo , a. masiero , and m. pietroni , phys . d * 74 * , 083505 ( 2006 ) . r. catena , n. fornengo , a. masiero , m. pietroni , and m. schelke , jhep * 10 * , 003 ( 2008 ) . m. kamionkowski and m.s . turner , phys . d * 42 * , 3310 ( 1990 ) . santiago , d. kalligas , and r.v . wagoner , phys . d * 58 * , 124005 ( 1998 ) . p. salati , phys . b * 571 * , 121 ( 2003 ) . g. lambiase , phys . d * 83 * , 107501 ( 2011 ) ; j. cosmol . ( 2012 ) 028 . s. capozziello , m. de laurentis , and g. lambiase , phys . b * 715 * , 1 ( 2012 ) . a. iorio , g. lambiase , eur . j. c * 75 * , 115 ( 2015 ) . g. gelmini and p. gondolo , arxiv:1009.3690 [ astro-ph.co ] . g. damico , m. kamionkowski , and k. sigurdson , arxiv:0907.1912 [ astro-ph.co ] . l. randal and r. sundrum , phys . 83 * , 4690 ( 1991 ) . f. profumo and p. ullio , jcap * 0311 * , 006 ( 2003 ) . f. rosati , phys . b * 570 * , 5 , ( 2003 ) ; phys . b * 571 * , 121 ( 2003 ) . c. pallis , jcap * 0510 * , 015 ( 2005 ) . r.catena , n. fornengo , a. masiero , m. pieroni , f. rosati , phys . d * 70 * , 063519 ( 2004 ) . e. komatsu _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 180 * , 330 ( 2009 ) .	by considering @xmath0 gravity models , the cosmic evolution is modified with respect to the standard @xmath1cdm scenario . in particular , the thermal history of particles results modified . in this paper , we derive the evolution of relics particles ( wimps ) assuming a reliable @xmath0 cosmological solution and taking into account observational constraints . the connection to the pamela experiment is also discussed . results are consistent with constraints coming from bicep2 and planck experiments .
direct and obvious method of light driven structuring is the using of a special light field configuration . generation of the light intensity grating and corresponding grating of excitations are widely discussed . more interesting and physically rich is structuring due to light driven self - organization . this ordering belongs to the class of self - organization phenomena in open dissipative systems and it is presented here . the typical examples of this class are benard convection @xcite , belousov - zhabotinsky reactions @xcite and turing instability @xcite . all self - organization phenomena are driven by some external flow through the system : heat in benard convection , chemical reagents in belousov - zhabotinsky reactions , etc . optical analogue of turing instability was presented in @xcite . in the latter case light amplitude has been modulated in time and space , which was crucial for the organization to occur . our investigations(see @xcite , @xcite , @xcite and references therein ) have shown that steady light flow through a system can provide ordering either . this light is a driving force which bunches the electrons in the case in hand . an electron in disordered media ( we shall call them `` glasses '' ) moves in random potential forming its energy spectrum . it consists of the extended electron and hole bands separated by the `` gap '' filled by the local states . we shall discuss here the case when photon energy is less than the gap in glass and light generates transitions between local states only . transitions between different traps are of a spacial interest . they are responsible for the light induced current and the structuring of the matter discussed . the electron transfer from an initial trap @xmath1 to the final trap @xmath2 may be treated as generation of electron and hole at sites @xmath2 and @xmath1 respectively . strong coulomb electron - hole interaction is changed considerably with the shift of the distance between traps @xmath2 and @xmath1 therefore strong electron - phonon coupling takes place . light absorption in the case of strong electron - phonon interaction is well studied . the rate @xmath3 of the electron transition between sites @xmath1 and @xmath2 is given by the formula @xcite @xmath4 where @xmath5 is photon flux , @xmath6 is photon energy , @xmath7 is the energy of the electron excitation ( difference between electron energies @xmath8 in trap @xmath2 and @xmath9 in trap @xmath1 ) ; @xmath10 ; @xmath11 , where vectors @xmath12 and @xmath13 points positions of the traps ; @xmath14 is the angle between @xmath15 and the light polarization vector , @xmath16 , @xmath17 , @xmath18 is the stokes shift . the light driven electron transitions in hand are accompanied by phonon generation . phonons are excluded from the formula ( [ w1 ] ) therefore exact energy conservation law @xmath19 ( @xmath20 is phonon energy ) looks like conservation with the accuracy @xmath21 : according to ( [ w1 ] ) an electron may be transfered to different levels in the range of the band width @xmath22 . absorption band ( [ w1 ] ) as a function of @xmath23 gains maximum ( resonance ) at the energy @xmath24 . we shall call the excited states with the @xmath25 as low energy excitations ( deep levels ) and excited states with the @xmath26 as high energy ones . as a first step we shall examine light driven kinetics of a single electron in external electric field @xmath27 . it is nontrivial : as we shall see electron is shifted by light predominantly in opposite to the electric force @xmath28 direction . the idea of such behavior is following . if vector of electron transfer is @xmath29 then dipole moment generated is @xmath30 ( @xmath31 is electron charge ) . field contribution to the energy of the excitation is @xmath32 . if the electron is transported in opposite to the electric force direction ( @xmath33 ) the energy of the electron excitation @xmath34 becomes larger ; low energy excitation becomes closer to resonance @xmath35 and therefore probability of its generation increases . this means that electron transfer to the deep levels goes predominantly in opposite to the electric force direction . electron transitions to high levels go mainly in the direction of the force but these levels have less life time in comparison with deep levels therefore contribution of the deep levels prevails . in the case of two electrons according to the above idea light would push each electron in the direction of the other electron . let the second electron is embedded at positions @xmath36 then coulomb interaction shifts all energy levels for the first electron and , hence , shifting the excitation energies by @xmath37 for the transition @xmath38 of the first electron . transitions which reduces electron - electron distance give positive contribution to the coulomb energy and shift deep excitations towards the resonance @xmath35 . that is why low energy transitions of an electron are directed preferably towards the other electron and effective electron - electron interaction becomes attractive under strong pumping . due to shorter life time high energy excitations play minor role in comparison with deep levels and therefore contribution of low energy excitations dominates . at high pumping these transitions dominate also over ordinary mobility which is quite low and independent on laser power . effective attraction between electrons makes their homogeneous distribution unstable and electron bunches are formed . phonon assistance designates two peculiar features of our system : first , it is dissipative ( not hamiltonian ) ; secondly , phonons brake phase relations for the electron wave functions therefore nondiagonal elements of the density matrix vanish and only probabilities of the site populations appear in consideration and they give full and correct description of the system . if the energy difference @xmath39 ( electron loses energy ) , there are two channels to transit : first , it can undergo light - induced transition with the rate being described in a way analogous to ( [ w1 ] ) with the only change @xmath40 to @xmath41 ( @xcite ) ; second , it may relax to a lower energy state with the rate @xmath42 . so , for the electron transition process with the energy loss we have : @xmath43 the factor @xmath44 is of the order of the inverse lifetime for excited electron states in an atom , so that @xmath45 . light driven kinetics is governed by dimensionless pumping constant @xmath46 . wave function of trapped electron outside the trap @xmath1 is @xmath47 , where @xmath48 , @xmath49 is the electron mass , @xmath50 is spacing of bottom of trapped levels from the bottom of the extended states . normal scale for this parameter is @xmath51 . wave function of deep levels vanishes rapidly outside the trap while shallow levels have long tails of wave functions . their overlap is determined by the slowest function , therefore @xmath52 ( @xcite ) . in the frame of the ( [ w1 ] ) , ( [ w2 ] ) master rules we studied light driven motion of two electrons and calculated their density - density correlation function @xmath53 where @xmath54 and @xmath55 are the occupation numbers of the corresponding traps , @xmath56 is time . we normalize the correlation function by the factor @xmath57 , where @xmath58 is dimensionality of the system . this normalization avails to see the interaction effect : if the electrons were not interacting , one would have @xmath59 for the noncorrelated case . the typical @xmath60 for three dimensions is shown in fig . [ fig2 ] . , @xmath61 , @xmath62 , electron density is constant within the range @xmath63.,width=604,height=529 ] the bar chart displays the probability of different electron - electron ranges . electron traps were randomly distributed in space and single electron level @xmath64 in each trap @xmath1 is also random according to some density of states . one electron in position @xmath36 shifts each initial level @xmath1 of the second electron to the energy of coulomb interaction ( and vice versa ) @xmath65 the parameters used for calculation are : average trap distance @xmath66 _ _ , @xmath67 , @xmath68 , impurity band width @xmath62 , @xmath69 , @xmath70 . the bound state of two electrons is formed at pumping constant @xmath71 . pumping constant @xmath72 corresponds to laser power @xmath73 . we found that at high pumping in wide region of parameters the correlation function has such a maximum . despite the coulomb repulsion two electrons prefer to stay at some finite distance . analogous behavior has been observed for @xmath74 and @xmath75 . increase of average distance @xmath76 between the traps ( decrease of coulomb interaction ) results in narrowing and vanishing of the peak in @xmath60 . at lower @xmath77 the peak in @xmath60 vanishes also ( see for the details @xcite ) . electron motion reveals competition between ordinary mobility in the direction of the electric force @xmath78 given by relaxation term @xmath79 in ( [ w2 ] ) and light driven electron transfer in opposite direction . at long distance between the electrons the light induced transitions dominate and an electron is pushed to the other electron : light overcomes coulomb repulsion ( ! ) . this motion is stopped at short distances where coulomb repulsion dominates : electrons do not penetrate to short distance where correlation function tends to zero ( fig . [ fig2 ] ) . electric field at this distance is @xmath80 and this is an estimation for the effective electric field caused by light induced electron transfer . this field is close to the damage field @xmath81 . as the result of the competition bound state of two electrons is formed with some preferable distance between particles . effective attraction between electrons results in light driven electron bunching . we examined behavior of few electrons or even @xmath82 electrons and found quite similar peculiarities . energy shift of the initial levels for hopping electron is analogous to ( [ epsilon ] ) but is given by all other electrons : @xmath83 is decreased under light pumping ( parameters are @xmath84 , @xmath85 _ _ , @xmath86 _ _ , @xmath87 , number of electrons @xmath88.).,width=604,height=529 ] , @xmath85 _ _ , @xmath86 _ _ , @xmath87 , @xmath89).,width=604,height=529 ] we studied temporal change of the average size of the electron cloud @xmath90 and coulomb energy @xmath91 . the result is shown in figs . [ fig7 ] , [ fig8 ] : electron cloud is pressed by light into compact bunch . we see again electron motion which is impossible in thermodynamic case : the system tends to the state with higher energy ( fig . [ fig8 ] ) . bunch formation is accompanied by strong increase of the electric field at the boundary of the bunch . the formation is stopped by this field when it gains the value @xmath92 . it is convenient to introduce ground state and electron - hole presentation for the investigation of many body system in glass . glass never attain full thermodynamic equilibrium but partial equilibrium is possible and one may consider that at ground state all electron levels bellow some energy @xmath93 are occupied and states above this level are empty . energy shift of a level is determined in this case by the contribution of all electrons and holes available . @xmath94 contribution to the energy of the first excited electron is given by its `` own '' hole only @xmath95 where @xmath36 is position of the hole . calculation of light induced polarization have shown that it is directed in opposite to an external field direction ( fig . [ fig3 ] ) and therefore amplifies this field . units vs dimensionless time @xmath96 for parameters @xmath97 _ _ , @xmath98 , external field @xmath99 , @xmath100 , @xmath101.,width=529,height=453 ] the studied three dimensional many body system with long range coulomb interaction reveals self - organization effect even at weak pumping @xmath102 . we examined behavior of many body system of different size for one- , two- and three - dimensional samples . in main features it is the same for all dimensions . the corresponding model is presented in details in @xcite , @xcite , @xcite . computer simulation of the electron kinetics in the finite sample under external electric field shows that in agreement with the above idea electrons are transported by light predominantly in the direction opposite to the electric force ( electric current flows contra the voltage drop ) . this transfer is in contrast to the electron motion in a thermodynamic case when electrons move in the direction of the external force reducing in accordance with the le chatelet principle the external field ( negative feedback ) . in our case glass reveals positive feedback in response to the electric field . as a result strong static polarization directed in opposite to the external electric field is established . this polarization amplifies the initial field considerably . at the field of saturation light induced electron transitions are in the dynamical equilibrium with the ordinary electron mobility in opposite direction . computer simulation of the light driven electron - hole kinetics have shown positive feedback in response to static electric field in wide region of parameters independently on the dimension of the system . this unusual response makes homogeneous electron distribution unstable : an electron moves towards another electron or towards a cluster of electrons and macroscopic bunches are formed . we performed special experiment in order to observe directly the discussed bunching . strong laser wave with photon energy @xmath103 was used . band gap in our glass @xmath104 and therefore direct generation of free electrons and holes is impossible . however in the region of strong electric field due to franz - keldysh effect @xcite , @xcite this process becomes effective . energy discrepancy @xmath105 between the glass gap and photon energy is gained from the static electric field in process of electron tunneling ( the corresponding estimation see in @xcite ) . so , domain boundaries due to strong electric field become a weak chain in glass bonds and light driven bond breaking ( local fusion ) reveals bunch structure of the self - organized state . high quality uv - grade fused silica samples ( ku-1 ) with thicknesses of @xmath106 have been used in the experiment . before the experiment the samples have been subjected to ultrasonic cleaning in acetone and ethanol followed by washing in de - ionized water . two beams from the unstable resonator of the @xmath107 excimer laser ( @xmath108 , cl-7000 , pic gpi ) were focused on the rear side of a sample by the lens made of @xmath109 . we changed the angle between beams and therefore produced grating with different periods at the output surface of the glass . @xmath107 excimer laser generated @xmath110 fwhm pulses with energies up to @xmath111 and pulse repetition rate up to @xmath112 ( @xmath113 ) . ablation has been performed in ambient air as well as in a hermetic chamber that was purged by nitrogen or evacuated . silica surface morphology and etched structures depth have been analyzed with an optical microscope with further recording with cohu-4812 ccd camera . photograph of the surface after the light treatment is presented in fig . [ fig13 ] . . spatial coherence length exceeds 10 mm.,width=529,height=453 ] ) of incoherent beam ( spatial coherence length is less then @xmath114 ; b ) surface obtained by highly coherent radiation treatment with spatial coherence length @xmath115 in all directions . structures with the same spatial period @xmath116 are formed at different conditions a ) , b ) , and fig . [ fig13].,width=604,height=453 ] one can see farrows ( grating ) corresponding to the light intensity grating with the period @xmath117 and self - organized structure of each of them with the period @xmath0 corresponding to the discussed here self - organized electron bunches . this size has nothing in common with the light field scale ( wavelength , beam size , coherent length etc . ) , it is determined by internal properties of the glass what is characteristic feature of self - organized system and coincides exactly with the period of the bubble structure found in single wave experiment shown in fig . [ fig12 ] @xcite . we changed coherent length of the light waves , beam sizes , beam power and found that surface profiles ( figs . [ fig13 ] , [ fig12 ] ) remain the same : all experiments reveals the same spatial period @xmath118 what is typical for the self - organized systems . the material sputtering ( ablation ) occurs on the rear side of a fused silica plate with no stimulation of the process ( such as plasma generation in the vicinity of a sample or specific uv absorbing medium @xcite ) . analogous behavior was observed in @xcite where surface structuring took place at the @xmath119 interface and was explained in another way as interface property . our theoretical and experimental study of light driven electron kinetics in silica glass have shown that homogeneous distribution of the trapped electrons becomes unstable under light pumping and macroscopic electron ( hole ) bunches are formed . in wide region of the external parameters light pushes electrons into micron size coagulums which we observed in ablation process . under different conditions structures with the same spatial period @xmath120 are formed . this new length is determined by the internal properties of the material and is idependent on light properties ( beam size , beam power , coherent length etc . ) it was found that uv beam @xmath121 provides bunching and subsequent local bond breaking effectively but wave @xmath122 can not drive this process . deeper electrons with short tails of electron wave functions are involved in this case ( @xmath123 is increased ) and probabilities of the transitions are less . in addition the energy discrepancy @xmath105 is increased therefore probability of franz - keldysh tunneling becomes negligable . we observed also red luminescence of the glass sample under uv pumping recently reported in number of papers ( see @xcite and cited in @xcite references ) . ordering under light pumping discussed here may be the basement of the so called `` compaction effect '' ( see @xcite and references therein ) being the serious problem in lithography . ordered by light glass state certainly is more compact . this work was performed under auspices of russian academy of sciences in the frame of the program `` optical spectroscopy and frequency standards '' and russian foundation for the basic research under grant 04 -02 - 16394 . 6 h. benard , rev . pures appl . * 12 * 1261 ( 1900 ) . zaikin , a.m. zhabotinsky , nature * 225 * 535 ( 1970 ) . a.m. turing , phil . london b * 237 * 37 ( 1952 ) . f.t . arecchi , physica d * 86 * 297 ( 1995 ) . b.p . antonyuk and v.b . antonyuk , physics - uspekhi * 44 * 53 ( 2001 ) . antonyuk , v.b . antonyuk , a.z . obidin , s.k . vartapetov and s.k . kurzanov opt . comm * 230 * 151 ( 2004 ) . boris p. antonyuk , `` light driven self - organization '' , nova science publishers , inc . , ny ( 2003 ) . r. kubo and y. toyazawa , progr . * 13 * 160 ( 1955 ) . franz , naturforsh . . 13a s. 484 ( 1958 ) . keldysh , jetp * 34 * 1138 ( 1958 ) . k. sugioka , j. zhang , k. midorikawa , laser machining method and laser machining apparatus , patent no . : us 6 180 915 b1 , ( 2001 ) . yu , j.v.zhang , i.w . boyd and y.f . lu , appl . a * 72 * 35 ( 2001 ) . r. schenker and w. oldham , j. vac . technol b * 14 * 3709 ( 1996 ) .	theoretical and experimental evidence of light driven structuring of glasses is presented . we show that light overcomes coulomb repulsion and effective electron - electron interaction in glasses under strong light pumping becomes attractive . as the result homogenious distribution of trapped electrons gets unstable and macroscopic electron bunches are grown . at different conditions ordered structures with period @xmath0 determined by internal properties of the material are formed these structures were observed in ablation : surface profile after laser treatment reveals ordered pattern corresponding to the light induced electron domains .
a classical technique for constructing quadratic number fields with class number divisible by @xmath5 is studying integral solutions of the equation @xmath7 for each integral point @xmath8 we can form the ideal @xmath9 in the ring of integers of @xmath10 ; the ideal @xmath11 has norm @xmath12 and satisfies @xmath13 , hence generates an ideal class of order dividing @xmath5 . it seems that p. joubert @xcite was the first to observe that a class of prime order @xmath5 in the group of binary quadratic forms with negative discriminant @xmath3 implies the solvability of the equation ( [ yamamoto ] ) ; joubert used techniques from the theory of complex multiplication for exploiting this observation . nagell @xcite later used ( [ yamamoto ] ) for proving the existence of infinitely many complex quadratic number fields with class number divisible by @xmath5 . by extending nagell s approach , yamamoto @xcite was able to prove the existence of infinitely many _ quadratic number fields with class number divisible by @xmath5 . the same approach was further extended by various authors ; we mention in particular craig @xcite . in this article , we interpret ( [ yamamoto ] ) as an affine surface and show that a certain subset @xmath14 of the integral points on ( [ yamamoto ] ) can be given a group structure in such a way that 1 . the integral points on the hyperplane @xmath15 , which lie on the pell conic @xmath16 , form a subgroup with respect to the classical group structure on pell conics ( see @xcite ) ; 2 . there is a surjective group homomorphism @xmath17 $ ] to the @xmath5-torsion of the narrow class group of @xmath18 . these results explain the success of yamamoto s approach , and at the same time raise a few new problems that we do not yet fully understand . the rational points on the surface lying on the hyperplane @xmath19 form ( the affine part of ) a hyperelliptic curve @xmath20 ; in the case @xmath21 , this is an elliptic curve . although the integral points on @xmath22 do not form a group in general , it was observed by buell @xcite and soleng @xcite that the integral points on @xmath22 ( and , more generally , certain rational points satisfying some technical conditions ) give ideal classes of order dividing @xmath23 in such a way that the map from @xmath22 to the @xmath23-class group respects the group law on the elliptic curve , i.e. , that collinear points get mapped to classes whose product is trivial . blling @xcite has extended these results to the hyperelliptic curves lying on the surface ( [ yamamoto ] ) . @xmath24{conic.jpg } & \includegraphics[width=5.0cm]{elliptic.jpg } \end{array}$ ] @xmath24{negconic.jpg } & \includegraphics[width=5.0cm]{negelliptic.jpg } \end{array}$ ] although we will see below that the group law is best understood by using ideals in quadratic number fields , the explicit addition formulas are tied closely to the composition of binary quadratic forms . for this reason , we replace the equation ( [ yamamoto ] ) of the surface by @xmath25 , where @xmath2 is the principal form with discriminant @xmath3 defined below . for a brief introduction to the composition of binary quadratic forms via bhargava s cubes see lemmermeyer @xcite ; more details along more classical lines can be found in flath @xcite . let @xmath3 be a fundamental discriminant ( the discriminant of a quadratic number field ) . the principal form with discriminant @xmath3 is defined by @xmath26 by @xmath27 we denote the binary quadratic form @xmath28 . such a form @xmath29 represents an integer @xmath30 if @xmath31 for some integers @xmath32 ; it is said to represent @xmath30 primitively if , in addition , @xmath33 . an integral point @xmath0 on the pell surface @xmath34 is called primitive if @xmath35 . the set of primitive points on @xmath36 will be denoted by @xmath14 . now consider eqn . ( [ yamamoto ] ) and map a point @xmath0 on the pell surface ( [ esn ] ) to a point @xmath8 on ( [ yamamoto ] ) by setting @xmath37 this clearly gives a bijection between the integral points on these surfaces . in addition , yamamoto s condition @xmath38 is easily seen to be equivalent to the primitivity of @xmath0 , that is , to @xmath35 . let @xmath39 denote the ring of integers of the quadratic number field @xmath40 . there is a natural map @xmath41 defined by @xmath42 , where @xmath43 and @xmath44 is defined by @xmath45 . the elements in the image of @xmath46 have the property that their norms are @xmath5-th powers : @xmath47 . consider the set @xmath48 of nonzero elements in @xmath39 and its subset @xmath49 of nonzero natural numbers . the set @xmath50 , using @xmath51 to refer to positive integers which are @xmath5-th powers , is a group with respect to multiplication : the neutral element is @xmath52 , the inverse of @xmath53 is @xmath54 ( the element @xmath55 is , up to sign , simply the conjugate @xmath56 of @xmath57 , and so belongs to @xmath48 ) . the norm map induces a group homomorphism @xmath58 , where @xmath59 and @xmath49 denote the monoids of nonzero and of positive integers , repectively . observe that if @xmath60 are primitive ( this means that @xmath61 for all primes @xmath62 ) and @xmath63 , then in general @xmath64 can not be chosen to be primitive . an example is provided by @xmath65 and @xmath66 , where @xmath67 ; here @xmath68 is not generated by a primitive element for any @xmath69 . on the other hand we shall prove below [ p1 ] the cosets of primitive elements in the kernel of the norm map @xmath58 form a subgroup @xmath70 of @xmath50 . this fact allows us to prove that there is a bijective map @xmath71 given by @xmath72 ; using this bijection we can make @xmath14 into an abelian group . the situation is summed up by the following diagram : @xmath73 [ t1 ] the map @xmath71 is bijective ; thus @xmath14 becomes an abelian group by transport of structure . injectivity : assume that there are elements @xmath74 with @xmath75 . then there exist @xmath76 with @xmath77 , and the primitivity of @xmath78 and @xmath79 implies that @xmath80 and @xmath81 must be units . since @xmath76 , this implies @xmath82 . surjectivity : assume that @xmath83 is primitive with @xmath84 . then @xmath85 for some number @xmath86 implies @xmath87 , hence @xmath88 with @xmath89 . observe that the neutral element of @xmath14 is the point @xmath90 , and that the inverse of @xmath0 is given by @xmath91 observe also that the integral points on the pell conic @xmath92 , which correspond to the points @xmath93 on the pell surface , inherit their classical group structure since @xmath94 , where @xmath95 if @xmath96 and @xmath97 if @xmath98 . in fact , since the elements @xmath99 have norm @xmath100 , the element @xmath101 is always primitive . for proving prop . [ p1 ] we use the following characterization of primitive elements : [ lprim ] let @xmath102 be a nonzero element of the order @xmath39 . 1 . @xmath57 is primitive if and only if @xmath103 for some ideal @xmath104 dividing the product of all ramified primes . if @xmath105 for some @xmath69 , then @xmath57 is primitive if and only if @xmath106 . assume first that @xmath57 is primitive , let @xmath107 be an unramified prime ideal , and set @xmath103 . if we had @xmath108 , then @xmath109 and @xmath110 . since @xmath107 is unramified , the prime @xmath111 below @xmath107 either splits ( and then @xmath112 ) , or @xmath113 is inert . in both cases we deduce that @xmath114 , which contradicts our assumption that @xmath57 be primitive . conversely , assume that @xmath104 divides the product of all ramified primes . if @xmath115 for some prime @xmath62 , then @xmath116 , hence @xmath117 . this shows that @xmath118 is divisible either by an unramified prime ideal or by the square of a ramified prime ideal . this proves the first statement . for proving b ) , assume first that @xmath106 ; then @xmath119 is primitive by what we have already proved . finally , if @xmath120 and @xmath57 is primitive , then @xmath104 is a product of ramified prime ideals . but if @xmath121 for some ramified prime ideal @xmath107 above @xmath111 , then @xmath122 , and this is impossible for @xmath69 . let @xmath57 be a primitive element . if @xmath123 , then @xmath124 is an @xmath5-th ideal power . the converse holds if @xmath57 is totally positive . the claim is trivial for @xmath125 ; assume therefore that @xmath69 . if @xmath57 is primitive and @xmath120 , lemma [ lprim ] implies that @xmath57 and @xmath56 are coprime . now @xmath126 implies that @xmath124 is an @xmath5-th ideal power . now assume that @xmath124 . then @xmath127 for some positive integer @xmath128 , and since @xmath57 is totally positive , we have @xmath129 . assume that @xmath57 and @xmath130 are primitive elements representing cosets in the kernel of the norm map . write @xmath131 with @xmath132 and with @xmath133 maximal . we have to show that @xmath64 is primitive . assume not ; then @xmath134 for some rational prime @xmath111 . since @xmath61 and @xmath135 ( by the primitivity of these elements ) , the prime @xmath111 can not be inert , and there is a prime ideal @xmath107 above @xmath111 with @xmath109 and @xmath136 . since @xmath57 and @xmath130 are @xmath5-th ideal powers , we must have @xmath137 and @xmath138 , and this implies @xmath139 . by the maximality of @xmath140 we must have @xmath141 , and this implies @xmath142 . thus @xmath70 is closed under multiplication ; since the inverse of @xmath53 is @xmath143 ( with the sign chosen in such a way that @xmath144 ) , the set @xmath70 forms a subgroup of @xmath145 . * the points with @xmath146 on @xmath36 form a subgroup of @xmath14 ; such points @xmath147 correspond to units @xmath148 , and the group law is induced by the usual multiplication of units . this shows that the group law on the pell conic @xmath149 coincides with the standard group law on these curves . we have already remarked that the set @xmath14 was used to extract information on the @xmath5-torsion of the class group @xmath150 of the quadratic number field @xmath18 . given a point @xmath88 , we know that @xmath83 is an @xmath5-th ideal power : @xmath124 . sending @xmath57 to the narrow ideal class of @xmath11 we get a map @xmath151 $ ] from @xmath14 to the group of ideal classes ( in the strict sense ) in @xmath152 whose order divides @xmath5 : the map @xmath153\ ] ] is a surjective group homomorphism . proving that @xmath154 is a group homomorphism is easy : let @xmath155 @xmath156 with @xmath157 , and put @xmath158 . then @xmath159 , and @xmath160 for some @xmath161 that differs from @xmath162 by the @xmath5-th power of some positive integer @xmath140 . this implies that @xmath163 , hence @xmath164 as claimed . for proving that @xmath154 is onto , consider the narrow ideal class @xmath165 \in { \operatorname{cl}}^+(k)[n]$ ] for some ideal @xmath11 coprime to the discriminant . then @xmath166 for some @xmath83 . we claim that we can choose @xmath11 in such a way that @xmath57 is primitive : in fact , let @xmath111 be a prime dividing @xmath167 and @xmath168 . if @xmath111 is inert , then @xmath169 , and replacing @xmath11 by @xmath170 does not change the ideal class . if @xmath112 is split , then we must have @xmath171 and @xmath172 , so again @xmath169 . since @xmath11 is coprime to the discriminant , ramified prime ideals do not divide @xmath11 . since @xmath119 is principal in the strict sense , we have @xmath173 ; writing @xmath83 we find @xmath88 as claimed . observe that @xmath174 , the subset of all @xmath88 with @xmath175 , forms a subgroup of @xmath14 , and that the proof above shows that the natural map @xmath176 $ ] is surjective . it is in general difficult to tell whether a point @xmath88 gives rise to an element of exact order @xmath5 or not , or more generally , whether two points generate independent elements . in the following , we shall briefly recall the criterion used by yamamoto . to this end , we introduce a natural homomorphisms between the groups @xmath14 : assume that @xmath177 ; then there is a group homomorphism @xmath178 in order to avoid a problematic case , we let @xmath179 denote the set of all primitive points @xmath0 such that @xmath180 ; equivalently , @xmath78 is primitive and not divisible by any ramified prime ideal . assume that @xmath181 . with @xmath83 we have @xmath182 ; setting @xmath183 , we find @xmath184 , hence @xmath185 . observe that @xmath186 is primitive if @xmath57 is , except possibly when @xmath187 and @xmath57 is divisible by a ramified prime . setting @xmath188 , we have @xmath189 . since the map @xmath190 sending @xmath0 to @xmath191 is compatible with the group structure ( in fact : if @xmath192 , then raising this equation to the @xmath193-th power shows that @xmath194 ) , the claim follows . as an example , consider the surface @xmath195 ; using the point @xmath196 on @xmath179 we find @xmath197 , which gives us the point @xmath198 . it is desirable to have criteria for deciding whether a point @xmath199 is actually a newpoint , i.e. , does not come from @xmath200 for some proper divisor @xmath201 of @xmath5 . assume that @xmath202 . if @xmath203 and @xmath204 for some odd prime @xmath111 , then @xmath205 for some @xmath206 implies that @xmath207 is a @xmath111-th power modulo @xmath208 for every prime @xmath209 . let @xmath83 and @xmath124 . if @xmath210 for some @xmath206 , then @xmath211 for @xmath212 and @xmath213 . thus @xmath214 is a @xmath111-th power . let @xmath208 be a prime dividing @xmath128 ; then @xmath215 splits in @xmath193 , and we have @xmath216 and @xmath217 . if @xmath96 , then @xmath218 , hence @xmath219 , @xmath220 , and so @xmath221 this implies @xmath222 as claimed . now assume that @xmath98 . then @xmath223 shows that @xmath224 ( since @xmath225 ) , hence @xmath226 . this criterion is not very strong ; it does not detect that the points @xmath227 or @xmath228 on @xmath229 are newpoints . on the other hand , @xmath230 must be a newpoint since @xmath231 is not a cube modulo @xmath232 . let us now make the group law on @xmath14 explicit by deriving addition formulas @xmath233 from the definition of the group law it is clear that such addition formulas must involve computations of greatest common divisors . the following lemma contains the technical part of the proof : [ lgcd ] for points @xmath234 , @xmath235 , we set @xmath158 . let @xmath236 ; then @xmath237 is an @xmath5-th ideal power , and with @xmath238 , we have @xmath239 conversely , the gcd on the left hand side of ( [ egcdab ] ) is an @xmath5-th power , and if ( [ egcdab ] ) holds , then @xmath240 for an ideal @xmath241 with norm @xmath242 . since @xmath159 , the ideal @xmath104 must be an @xmath5-th power . from @xmath243 and @xmath244 we deduce that @xmath245 , and now @xmath246 implies @xmath247 . if , conversely , @xmath111 is a prime dividing @xmath248 , then the primitivity of @xmath249 implies that @xmath112 must be split in @xmath18 . if , say , @xmath250 , then the primitivity of @xmath251 shows that we must have @xmath252 and therefore @xmath253 . thus if @xmath254 is the exact power of @xmath111 dividing @xmath30 , then @xmath255 is the exact power of @xmath107 dividing @xmath251 , and the fact that @xmath256 is an @xmath5-th ideal power shows that @xmath201 must be a multiple of @xmath5 . this implies that * @xmath257 must be an @xmath5-th power , * @xmath258 is the norm of an ideal @xmath241 , * and @xmath259 . this completes the proof . now we can present the explicit formulas for adding points on @xmath14 : [ pexpl ] for @xmath260 we have the addition formula ( [ eabc ] ) , where @xmath261 with @xmath242 as in ( [ egcdab ] ) . the group law is defined via @xmath262 , where @xmath161 is required to be primitive . equation ( [ eal12 ] ) and lemma [ lgcd ] show that @xmath263 . taking norms yields @xmath264 , and this proves the claim . on @xmath265,width=207 ] since there is a bijection between ideal classes and equivalence classes of binary quadratic forms , we can also describe the group law in terms of forms . it turns out that the geometric aspects of the description of the group law on @xmath14 in terms of forms adds a lot to our understanding of the arithmetic of pell surfaces and pell conics . for this reason , we will now construct a map sending primitive points on @xmath14 to primitive quadratic forms with discriminant @xmath3 . given @xmath88 , consider the form @xmath266 in order to get positive definite forms if @xmath267 we now agree to replace @xmath14 by @xmath268 in this case . it is easily checked that @xmath269 ; moreover , dirichlet composition immediately shows that @xmath270 is the principal form ( with discriminant @xmath271 ) . for constructing a form with discriminant @xmath3 , we have to `` underive '' @xmath272 . this process replaces a form @xmath273 with discriminant @xmath271 by an equivalent form @xmath274 , and then maps it to @xmath275 , which is a primitive form with discriminant @xmath3 . mapping @xmath199 to the equivalence class of the form @xmath276 turns out to be a homomorphism @xmath277 $ ] . underiving @xmath272 is accomplished by changing the middle coefficient modulo @xmath278 in such a way that it becomes a multiple of @xmath168 . for motivating the following lemma , consider the equation @xmath279 ; dividing through by @xmath280 and reducing mod @xmath128 yields @xmath281 , and this congruence has a unique solution . in this way we find given a point @xmath203 with @xmath282 , let @xmath130 be an integer satisfying the congruence @xmath283 ; then @xmath284 . define a quadratic form @xmath285 with @xmath286 . then @xmath276 is a primitive form with discriminant @xmath3 , and @xmath276 is positive definite if @xmath267 . assume now that @xmath96 , and set @xmath288 . from @xmath283 we see that there is an integer @xmath193 with @xmath289 . setting @xmath290 we find @xmath291 with @xmath292 ; the integer @xmath293 is determined by @xmath294 , which gives @xmath295 . setting @xmath296 , the form @xmath297 is primitive , has discriminant @xmath3 , and the fact that @xmath175 implies that @xmath298 is positive definite if @xmath267 . sending @xmath300 defines a map @xmath301 between two abelian groups ; we already know that the corresponding map to the ideal class group is a homomorphism , and of course the same holds for form classes . we will check the details below ; now let us determine the kernel of @xmath302 . to this end , recall how we constructed @xmath302 : to a point @xmath88 we have attached a quadratic form @xmath303 with discriminant @xmath271 ; this form @xmath272 is equivalent to a form @xmath304 , and underiving @xmath305 gave us @xmath285 . now a point @xmath203 is in the kernel if and only if @xmath306 , which happens if and only if @xmath276 represents @xmath100 . multiplying @xmath307 through by @xmath308 this shows that @xmath309 ( conversely , this equation implies @xmath307 ) . but @xmath305 represents @xmath308 properly if and only if the equivalent form @xmath272 does . thus we have shown for deciding whether the point @xmath227 on @xmath229 is in the kernel of @xmath302 we have to look at @xmath313 . this equation has solutions if and only if the form @xmath314 with discriminant @xmath315 represents @xmath100 , hence is equivalent to the principal form @xmath2 . this is not the case , since @xmath316 . we may also multiply the original equation through by @xmath317 and complete squares ; this gives @xmath318 . this equation is clearly unsolvable in integers , but has rational solutions , such as @xmath319 , for example ; this implies that we do not have a chance to show the unsolvability of the equation using congruences or @xmath111-adic methods . [ [ the - map - b - mathcal - s_nmathbbz - longrightarrowoperatornamecldelta - is - a - homomorphism . ] ] the map @xmath301 is a homomorphism . ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ consider a point @xmath320 on @xmath14 . we know that @xmath321 for some ideal @xmath11 in the maximal order of @xmath152 . for finding the form attached to @xmath11 we have to find an oriented @xmath322-basis @xmath323 of @xmath11 . let @xmath154 be an integer such that @xmath324 ; then @xmath325 , where @xmath130 denotes an integer in the residue class @xmath326 . it is easy to see that @xmath327 has the desired properties ; the form attached to @xmath11 then is @xmath328 where @xmath329 . in particular , @xmath330 . r. blling , _ ber einen homomorphismus der rationalen punkte elliptischer kurven _ , math . * 96 * ( 1980 ) , 207244 d. buell , _ class groups of quadratic fields _ , math . comp . * 30 * ( 1976 ) , d. buell , _ elliptic curves and class groups of quadratic fields _ , j. london math . ( 2 ) * 15 * ( 1977 ) , 1925 m. craig , _ a type of class group for imaginary quadratic fields _ , acta arith . * 22 * ( 1973 ) , 449459 d. flath , _ introduction to number theory _ , wiley & sons , new york , 1989 p. joubert , _ sur la thorie des fonctions elliptiques et son application a la thorie des nombres _ , c.r paris * 50 * ( 1860 ) , 774779 f. lemmermeyer , _ kreise und quadrate modulo @xmath111 _ , math . sem . * 47 * ( 2000 ) , 5173 f. lemmermeyer , _ conics a poor man s elliptic curves _ , arxiv : math/0311306v1 , ( 2003 ) f. lemmermeyer , _ higher descent on pell conics iii the first 2-descent _ , + arxiv : math/0311310v1 , ( 2003 ) f. lemmermeyer , _ binary quadratic forms and counterexamples to hasse s local - global principle _ , preprint 2009 t. nagell , _ ber die klassenzahl imaginr - quadratischer zahlkrper _ , hamburg , * 1 * ( 1922 ) , 140150 r. soleng , _ homomorphisms from the group of rational points on elliptic curves to class groups of quadratic number fields _ , j. number theory * 46 * ( 1994 ) , 214229 y. yamamoto , _ on unramified galois extensions of quadratic number fields _ , osaka j. math . * 7 * ( 1970 ) , 5776	we define a group stucture on the primitive integer points @xmath0 of the algebraic variety @xmath1 , where @xmath2 is the principal binary quadratic form of fundamental discriminant @xmath3 and @xmath4 is fixed . a surjective homomorphism is given from this group to the @xmath5-torsion subgroup of the narrow ideal class group of the quadratic number field @xmath6 .
a magnetic configuration , previously in equilibrium in the solar atmosphere , can reach a global instability threshold when the magnetic stress becomes too high . in this case , the plasma is ejected into the interplanetary ( ip ) medium and is observed as a coronal mass ejection ( cme ) by the solar coronagraphs . this magnetized mass , which can be expelled as fast as few times @xmath2 km / s , is recognized in the ip space as an interplanetary cme , icme , see e.g. , . during its travel from the sun to 1 au , fast cmes are slowed down due to drag forces between the icme and the solar wind environment ( see , e.g. , ) . thus , at 1 au they can reach speeds as high as @xmath3 km / s . magnetic clouds ( mcs ) are a particular subset of icmes . they are formed by twisted magnetic flux tubes that carry a large amount of magnetic helicity from the sun to the ip medium . they also transport significant amounts of magnetic flux , mass , and energy . the principal characteristics of these magnetic structures are : ( i ) an enhanced magnetic field , ( ii ) a smooth rotation of the magnetic field vector through a large angle ( @xmath4 ) , and ( iii ) a low proton temperature @xcite . the magnetic field in mcs is relatively well modeled by the so - called lundquist s model @xcite , which considers a static and axially - symmetric linear force - free magnetic configuration ( e.g. , ; ; ; ; ) . however , many other different models have been also used to describe the magnetic structure of mcs . some models consider the mc as a rigid body during the time it travels through the solar wind and crosses the spacecraft . considered a cylindrical shape for the cloud cross - section and a non - linear force - free field ; while , , and considered a cylindrical cloud but a non - force free field . non cylindrical static models have been also applied to mcs ( e.g. , , ) . a comparison of global quantities ( magnetic fluxes and helicity ) derived from different static models has been done by and . different techniques have been compared using synthetic data and analyzing the output of numerical simulations of mcs @xcite . some mcs present a significantly larger velocity in their front than in their back , a characteristic of expansion . thus , some authors have used dynamical models to describe these clouds during the observation time ; they have considered two cases : only with a radial expansion ( see , e.g. , , , , ) , and with expansion in the radial and axial directions ( see , e.g. , , ) . some dynamical models consider an expanding elliptical shape for the mc ( e.g. , ) . the main aim of these models is to take into account the evolution of the magnetic field as the spacecraft crosses the mc ; then , to correct the effect of mixing spatial - variation / time - evolution in the observations to get a better determination of the mc field ( and related characteristics ) . we analyze the mc detected inside the icme observed at l1 between nov . 9 , 2004 , at 20:25 ut and nov . 11 , 2004 , at 18:45 ut @xcite . this is a very fast , left - handed and huge mc with a size larger than 0.2 au in the earth - sun direction . it has a very intense magnetic field ( @xmath5 nt ) , and expands strongly , with a velocity of @xmath6 km / s in its front and @xmath7 km / s in its back , depending on where the rear boundary is set . this mc presents one of the largest ( ever observed ) velocity differences between its front and its back @xcite . we analyze this mc using a model - independent method , called direct method @xcite . it takes into account the magnetic flux conservation in closed structures , such as flux ropes . this method gives us an estimation of the magnetic flux in the mc directly from the data , and allows us to improve the determination of the orientation of the mc axis , as well as its boundaries . finding the boundaries for some mcs is an open issue @xcite . indeed , several authors , using different proxies , set them at different times with the consequent differences in the estimated mc axial orientation and in the estimations of global magnetohydrodynamic quantities . in section [ section_mc ] , we describe the magnetic and plasma properties of the mc . then , in section [ section_direct_method ] , we present the direct method and the results obtained . in section [ sect_models ] , we fit the two dynamical models to velocity and magnetic field observations . from the fitted models we compute the magnetic fluxes , and compare them with the values obtained from the static lundquist s model and from the direct method . finally , in section [ conclusion ] , we give a summary and our conclusions . + in this section we analyze _ in situ _ magnetic and plasma observations of the mc observed on nov . 9 - 10 , 2004 , which is located inside an icme . we define a local frame of coordinates , attached to the mc , and we analyze the data in this frame . an icme was observed from 20:25 ut on 9 nov . to @xmath8 18:45 ut on 11 nov . the icme is preceded by a typical piled up solar wind material ( panel @xmath9 in figure [ fig_bv_gse ] ) . this corresponds to plasma and magnetic field pushed from behind by the icme , forming the turbulent pre - icme sheath ( notice the high level of fluctuations in @xmath10 ) . a forward shock is located in front of the sheath ( at 9:15 ut of nov . this event presents also a non - typical second shock inside the sheath that precedes the icme ( at 18:20 ut of nov . 9 ) . for a deeper description of shocks and their association with solar sources see , in particular their table i. the end of the icme is marked by a thick solid line ( nov . 11 , 18:45 ut , label `` end '' in figure [ fig_bv_gse ] ) as defined by . . 11 at 18:45 ut , the magnetic field is consistent with parker s spiral ( changes of magnetic sectors are observed in @xmath11 ) . the higher level of magnetic fluctuations , typical of the fast solar wind , confirms this interpretation . two mcs were initially reported inside this icme : the first one starting on nov . 09 , at 20:54 ut , and finishing on nov . 10 , at 03:24 ut , and the second one starting on nov . 10 , at 03:36 ut , and finishing on nov . 10 at 11:06 ut ( ` http://lepmfi.gsfc.nasa.gov/mfi/mag_cloud_s1.html ` ) . the rotation of the field is indeed larger than usual ( close to one turn , figure [ fig_bv_gse ] ) . moreover , the magnetic field observed has indeed a non - classical structure : it is very strong in the front , progressively decreases in the back and has an extended weak tail . still , a coherent and continuous variation of both the field strength and the plasma velocity is observed , without clear evidence of two independent magnetic structures . ) , latitude ( @xmath10 ) and longitude ( @xmath11 ) angles of @xmath12 in gse , the bulk velocity ( @xmath13 ) , the proton density ( @xmath9 ) , and the proton plasma beta ( @xmath14 ) , all of them as a function of time . the vertical dashed line `` in '' corresponds to the start of the field coherent rotation ( beginning of the mc , nov . 9 at 20:30 ut ) , the vertical dotted line to the cloud center ( nov . 10 at 02:02 ut , see section [ direct_method_structure ] ) , the vertical dash - dotted lines `` out@xmath15 '' and `` out@xmath16 '' to the extremes of the range of possible endings for the rotation of @xmath12 ( end of the cloud , nov . 10 at 08:15 ut and 10:20 ut ) , and the dash - dotted line `` out@xmath17 '' to a strong discontinuity ( in between this range , on nov . 10 , at 09:00 ut ) . the thin solid vertical line `` back '' marks a strong magnetic discontinuity ( nov . 11 , at 06:02 ut ) and the thick vertical solid line `` end '' the end of the icme ( nov . 11 , at 18:45 ut ) . horizontal dotted lines in @xmath10 and @xmath11 panels indicate the orientation of parker s spiral , while the one in @xmath14 panel marks the mean value for a set of mcs studied by lepping _ ( 2003 ) . ] in out@xmath15 out@xmath16 back end out@xmath17 center other studies concluded that only one extended mc was present . in and the boundaries of the magnetic cloud were selected as starting on nov . 9 , at 20:30 ut , and finishing on nov . 10 , at 10:00 ut . similar boundaries ( starting on nov . 9 , at 20:40 ut , and finishing on nov . 10 , at 10:20 ut ) were chosen by . both the direct method and data modeling confirm the presence of only one flux rope ( sections [ section_direct_method ] and [ sect_models ] ) . we analyze the _ in situ _ measurements of the magnetic field components obtained by the magnetic field instrument ( mfi , ) and the plasma quantities obtained by the solar wind experiment ( swe , ) , both aboard wind . there is a small data gap in swe from nov . 9 , 21:58 ut to 22:27 ut ( see figure [ fig_bv_gse ] ) . the magnetic field observations are in gse ( geocentric solar ecliptic ) coordinates . in this right - handed system of coordinates , @xmath18 corresponds to the earth - sun direction , @xmath19 points to the north ( perpendicular to the ecliptic plane ) and @xmath20 is in the ecliptic plane and points to the dusk when an observer is near earth ( thus , opposing to the planetary motion ) . the front boundary of the mc ( nov . 9 , at 20:30 ut ) is well defined ( vertical dashed line in figure [ fig_bv_gse ] , label `` in '' ) . the magnetic field presents a north - west - south rotation with time ( see @xmath10 and @xmath11 panels in figure [ fig_bv_gse ] ) ; thus , the mc is formed by a left - handed flux rope with its axis almost on the ecliptic and pointing to the west ( @xmath21 or @xmath22 ) . a characteristic of this mc is its very strong expansion ( see panel @xmath13 in figure [ fig_bv_gse ] ) . the observed plasma velocity , @xmath13 , goes from @xmath6 km / s at the beginning to @xmath23 km / s close to the cloud end , a difference of @xmath24 km / s in the observed time range ( @xmath25 hours ) . this implies an expansion of @xmath26 km / s per hour . the mc rear boundary is uncertain , the vertical dash - doted lines in figure [ fig_bv_gse ] indicate a possible range ( from nov . 10 at 8:15 ut to 10:20 ut , labels `` out@xmath15 '' and `` out@xmath16 '' , respectively ) . however , the decrease of @xmath27 and the strong expansion are still present at later times indicating a backward extension of the mc . the mean value of @xmath14 ( ratio of the proton pressure to the magnetic pressure ) is 0.12 for a sample of mcs studied by ( horizontal dotted line in panel @xmath14 of figure [ fig_bv_gse ] ) . then , another indication that the mc is more extended in the back is the presence of @xmath28 after 10:20 ut . this region extends up to a strong discontinuity in @xmath10 and density ( on nov . 11 at 06:02 ut , label `` back '' in figure [ fig_bv_gse ] ) . we call the region between `` out@xmath16 '' and `` back '' simply the back region of the mc . finally , there is a region with weak , but coherent , field between the `` back '' and `` end '' boundaries . the physical origin of these regions is analyzed in section [ direct_method_icme_structure ] . to facilitate the understanding of the mc properties , we define a system of coordinates linked to the cloud in which @xmath29 is along the cloud axis ( with @xmath30 ) . we define the latitude angle ( @xmath31 ) between the ecliptic plane and the cloud axis , as well as the longitude angle ( @xmath32 ) between the projection of the axis on the ecliptic plane and the earth - sun direction ( @xmath18 ) measured counterclockwise . then , when @xmath31=90@xmath33 ( @xmath31=-90@xmath33 ) the cloud axis is parallel ( antiparallel ) to @xmath19 and it points to the ecliptic north ( south ) . when @xmath31=0@xmath33 the cloud axis is on the ecliptic plane , @xmath32=0@xmath33 being the case of the axial field pointing toward the sun , and @xmath32=90@xmath33 ( @xmath32=270@xmath33 ) when it points to the terrestrial dusk ( dawn ) . since the speed of an mc is nearly in the sun - earth direction and it is much larger than the spacecraft speed ( which can be supposed to be at rest during the cloud observing time ) , we assume a rectilinear spacecraft trajectory in the cloud frame . the trajectory defines a direction @xmath34 ( pointing toward the sun ) ; then , we define @xmath35 in the direction @xmath36 and @xmath37 completes the right - handed orthonormal base ( @xmath38 ) . we also define the impact parameter , @xmath39 , as the minimum distance from the spacecraft to the cloud axis . then , we construct a rotation matrix from the gse system to the cloud system and obtain the components of the observed magnetic field in the cloud coordinates : @xmath40 , @xmath41 , @xmath42 . the local system of coordinates is especially useful when @xmath39 is small compared to the mc radius ( @xmath43 ) . in particular , for @xmath44 and an mc described by a cylindrical magnetic configuration @xmath45 , we have @xmath46 and @xmath47 when the spacecraft leaves the cloud . in this particular case , the magnetic field data will show : @xmath48 , a large and coherent variation of @xmath49 ( with a change of sign ) , and an intermediate and coherent variation of @xmath50 , from low values at one cloud edge , taking the largest value at its axis and returning to low values at the other edge . the minimum variance ( mv ) method @xcite has been used to estimate the orientation of mcs ( see e.g. , , , , , ) . it gives a good estimation if @xmath39 is small compared to @xmath43 and if the in-/out - bound magnetic fields are not significantly asymmetric . however , when the cloud presents a strong expansion , as in the event studied here , the directions derived by the mv method will mix two different effects in the variance of @xmath12 : ( 1 ) the effect of the coherent rotation of @xmath12 ( which provides the cloud orientation ) , and ( 2 ) the effect of the cloud aging ( the decrease of the field strength with time due to magnetic flux conservation combined with cloud expansion ) . this latter effect is not associated with the cloud orientation ; thus , we apply the mv technique to @xmath51 to decrease the cloud aging consequences . we start the analysis taking the `` in '' and the `` out@xmath16 '' boundaries , since the strongest magnetic discontinuity is located at `` out@xmath16 '' . with these boundaries , we find the typical shape of the components of @xmath12 in the cloud frame , as discussed above ( see section [ direct_method_structure ] for further justifications ) . the mv method applied to the normalized field gives @xmath31=-23@xmath33 and @xmath32=274@xmath33 . with the same procedure and an end boundary on nov . 10 , 10:00ut , found @xmath31=-20@xmath33 and @xmath32=276@xmath33 . changing the cloud end between `` out@xmath15 '' and `` out@xmath16 '' , the ranges for @xmath31 and @xmath32 are : @xmath52[-25,0]@xmath33 and @xmath53[260,280]@xmath33 . is along the cloud axis ( @xmath54 and @xmath55 in gse coordinates ) , @xmath35 is orthogonal to both the mc axis and the spacecraft trajectory , and @xmath56 completes the right - handed orthogonal base . vertical lines correspond to the same times as in figure [ fig_bv_gse ] . the region between `` in '' and `` center '' is called the in - bound , and between `` center '' and `` out@xmath15 '' or `` out@xmath17 '' or `` out@xmath16 '' the out - bound . from `` out@xmath15 '' or `` out@xmath17 '' or `` out@xmath16 '' up to `` back '' we have the back region , while after that and up to `` end '' completes the icme extension . ] in out@xmath15 out@xmath16 back end out@xmath17 center figure [ fig_bv_cloud ] shows the components of the magnetic and velocity field in the cloud frame for an orientation of the cloud axis such that : @xmath54 and @xmath55 ( this orientation is justified in section [ direct_method_orientation ] ) . the magnetic field components show the typical large scale shape of mcs when the impact parameter is small compared with the cloud radius . however , an inner and non - typical sub - structure is present at its center ( mainly observed in @xmath57 , where it is antisymmetric ) . a strong expansion is observed in the velocity components , mainly along @xmath56 . this is expected from the mc orientation and the data in gse since , as @xmath58 , the spacecraft can not observe different parcels of fluid along a large range of @xmath59 values . however , a weak signature of expansion along the cloud axis can be observed in @xmath60 ( figure [ fig_bv_cloud ] ) . for an mc with @xmath61 and @xmath62 , an axial expansion is characterized by @xmath63 in the front , changing to @xmath64 in the back . this is just what is observed in our case when the mean value of @xmath60 is removed within the cloud . this gives us the clue that @xmath32 is larger than 270@xmath33 . when the spacecraft crosses a cylindrical mc ( or an elliptical one with one of the main axis parallel to the sun - earth direction ) and @xmath44 , @xmath65 . the first panel in figure [ fig_bv_cloud ] shows that the observed @xmath66 has a slightly negative mean value . the sign of @xmath66 , together with the evolution of @xmath57 , implies that the flux rope axis is above the ecliptic plane . in this section we summarize and extend the direct method presented by . this method lets us find the rear boundary of a flux rope for a given axis orientation , or the reverse , the mc orientation for a given position of the rear boundary . the front boundary of a flux rope is usually well defined by a discontinuity of the magnetic field ( changing abruptly from a fluctuating field in the mc sheath to a strong and coherent field within the mc ) . the corresponding current sheet is expected to be present all around the flux rope as the limit between magnetic regions with different magnetic connectivities ; so , with different magnetic stresses . a corresponding magnetic discontinuity , labeled `` out '' , is then expected at the rear of the mc . the flux rope is present in between these two discontinuities and the same amount of azimuthal flux is traversed twice by the spacecraft . let us consider a flux rope at a given time . the conservation of the magnetic flux ( @xmath67 ) across a plane formed by the spacecraft trajectory and @xmath68 ( @xmath69 constant ) gives @xcite : @xmath70 with @xmath71 being the spatial coordinates in the @xmath72 and @xmath68 directions , respectively . the observations provide only @xmath41 as a function of time along the trajectory . so , we need two hypothesis : an invariance of @xmath41 along the flux rope axis and the conservation of the magnetic flux with time . the first hypothesis is justified by a low ratio of the mc radius over the expected curvature radius of the axis and the balance of magnetic torques , which is expected to homogenize the field along the axial direction @xcite . the second hypothesis is valid as far as the amount of magnetic flux reconnected during the spacecraft crossing is low . indeed , we have found that magnetic flux is reconnected in the front of the mc ( see section [ direct_method_icme_structure ] ) . an estimation of the amount of flux reconnected during the spacecraft crossing gives @xmath73 5% of the initial azimuthal flux ( see section [ direct_method_reconnection ] ) ; so , unless the reconnection rate is much higher during the observing time than it was during the travel from the sun , the amount of reconnected flux is small during the crossing . we neglect below such reconnected flux . the elementary flux crossed during @xmath74 is @xmath75 , where @xmath76 is the axial length of the portion of the flux rope which had a length @xmath77 when the spacecraft entered the mc . then , equation ( [ byflux ] ) becomes : @xmath78 if the axis orientation and the position of one boundary of an mc are known , the above flux balance property can be used to find the mc center and the other boundary as follows . we define the accumulative flux per unit length : @xmath79 where @xmath80 is the time of the mc front boundary ( located at @xmath81 ) and @xmath82 . the position where @xmath83 has its absolute extreme gives an estimation of the position where the spacecraft reaches the closest distance to the mc axis . this indicates the @xmath84 position of the mc center , being this estimation more precise as the impact parameter is lower . then , when @xmath85 goes back to zero at @xmath86 , we have the other boundary . the region from @xmath81 to @xmath86 defines the mc flux rope . since the ratio of the mc radius to the sun distance is small ( typically @xmath87 or lower ) , @xmath88 . below we derive a correction to this estimation . if the mc axis does not change drastically its shape , and if it does not disconnect from the sun during the crossing time , its length is evolving proportionally to its distance to the sun , @xmath89 , so @xmath90 ( where @xmath91 au for the present observations ) . moreover , the cloud global velocity , which is the velocity of its center @xmath92 , is not expected to change significantly during the crossing time ; then , we have : @xmath93 all terms in the right hand side of equation ( [ byaccumul ] ) can be derived from observations . perpendicular to both the trajectory of the spacecraft and to the cloud axis ( @xmath57 , thin line curve ) and accumulated magnetic flux of this component per unit length ( @xmath94 , thick line curve ) . the dotted vertical line marks the cloud center ( nov . 10 , 02:02 ut ) . left and right panels correspond to computation of @xmath94 without ( @xmath95 ) and with axial expansion ( equation [ length_evol ] ) , respectively . in both panels the cloud orientation was taken as @xmath54 and @xmath55 . the vertical lines indicate the same positions as in figures [ fig_bv_gse ] and [ fig_bv_cloud ] . , title="fig : " ] perpendicular to both the trajectory of the spacecraft and to the cloud axis ( @xmath57 , thin line curve ) and accumulated magnetic flux of this component per unit length ( @xmath94 , thick line curve ) . the dotted vertical line marks the cloud center ( nov . 10 , 02:02 ut ) . left and right panels correspond to computation of @xmath94 without ( @xmath95 ) and with axial expansion ( equation [ length_evol ] ) , respectively . in both panels the cloud orientation was taken as @xmath54 and @xmath55 . the vertical lines indicate the same positions as in figures [ fig_bv_gse ] and [ fig_bv_cloud ] . , title="fig : " ] the direct method was previously applied to the oct . 18 - 20 , 1995 , mc @xcite . the orientation of its axis was well determined . the strong frontal discontinuity in the magnetic field was naturally related to another strong backward discontinuity , the flux balance given by equation ( [ byflux2 ] ) was satisfied , and @xmath76 was constant . the orientation of the axis of the mc studied here is not so well determined ; then , we explore different orientations to find which angles ( @xmath31 and @xmath32 ) give a cancellation of @xmath96 at the strongest magnetic discontinuity observed at the mc rear ( labeled `` out@xmath97 '' in figures [ fig_bv_gse ] , [ fig_bv_cloud ] ) , at the inner extreme labeled `` out@xmath98 '' , and at the intermediate time `` out@xmath99 '' ( between `` out@xmath98 '' and `` out@xmath97 '' ) . due to the orientation of the mc ( i.e. , its axis almost lying on the ecliptic and perpendicular to the sun - earth direction ) @xmath96 is mostly affected by the value of @xmath31 . we find @xmath100 from variations of the end boundaries in the full range between `` out@xmath98 '' and `` out@xmath97 '' , and using the two extreme possibilities on the axial expansion ( no axial expansion , @xmath101 , and axial expansion proportional to the distance to the sun , equation [ length_evol ] ) . the value of @xmath32 is constrained by imposing that @xmath102 should have a small variation with time ( as expected in flux rope models ) ; so , with no contribution of azimuthal or axial field components . this gives @xmath103 . left panel of figure [ fig_fy ] shows @xmath96 for @xmath101 ( no axial expansion ) and an orientation such that @xmath96 is canceled at the discontinuity `` out@xmath97 '' , which gives @xmath104 and @xmath105 . the rear boundary of the mc is in fact ambiguous from the data , since there is also a strong discontinuity between `` out@xmath98 '' and `` out@xmath97 '' ( `` out@xmath99 '' , at nov 10 , at 09:00ut ) , see figures [ fig_bv_gse ] and [ fig_bv_cloud ] . fixing the previous orientation , but using the axial expansion given in equation ( [ length_evol ] ) ( see panel in figure [ fig_fy ] ) , we find that the cancellation of @xmath96 is now at `` out@xmath99 '' . with this boundary , we find a much less extended region with a reversal of @xmath50 at the back of the mc . so , this boundary gives an mc field closer to the one inferred from a classical mc model . in this case , the results of the direct method agree with those of the fitted models ( including expansion , section [ sect_models ] ) . these are evidences that we have identified the right end boundary and orientation of the flux rope . component of @xmath12 in function of the accumulated magnetic flux of this component per unit length ( @xmath94 ) . @xmath94 is computed as in figure [ fig_fy ] . the difference in the amplitude of @xmath57 between the in-/out - bound regions is due to the mc aging ( not corrected here).,title="fig : " ] component of @xmath12 in function of the accumulated magnetic flux of this component per unit length ( @xmath94 ) . @xmath94 is computed as in figure [ fig_fy ] . the difference in the amplitude of @xmath57 between the in-/out - bound regions is due to the mc aging ( not corrected here).,title="fig : " ] the field component @xmath41 vanishes at three locations near the cloud center ( close to the vertical dotted line at the abscissa @xmath73 0.1 au in figure [ fig_fy ] ) . taking into account the expected antisymmetry of the in- and out - bound regions , we set the cloud center at nov . 10 at 02:02 ut ( dotted line in figure [ fig_fy ] ) . then , the accumulated flux @xmath94 ( equation [ byaccumul ] ) gives a unique relationship between the in- and out - bound data , since it labels each flux surface . this relationship is better shown using @xmath94 in the abscissa and reversing the sign of @xmath57 inside the out - bound branch ( figure [ fig_by_vs_fy ] ) . then , peaks and valleys of @xmath41 in the in-/out - bound branches can be related . as expected , this association is stronger near the mc center where the regions that are crossed are closer and also more isolated from the interaction with the solar wind environment @xcite . except for the strong discontinuity at the mc borders , the association between structures is not clear outside the core ( where @xmath41 has no characteristic variations that can be recognized in both the in- and out - bound branches ) . evolution along the spacecraft trajectory ( low left figure , a zoom of figure [ fig_fy ] ) . the magnetic structure of the flux rope is undulated , probably due to its fast evolution and interaction with the surrounding medium . the undulations are amplified in this scheme for clarity , but in the observed mc they are significant enough to produce reversals of @xmath41 . dot at cc indicates the location of the cloud center . numbers 1 , 2 , 3 , 4 and 5 are reference points of the observed field lines . ] there is a clearly distinguishable sub - structure in the cloud center seen as reversed peaks in @xmath41 and as a valley in @xmath94 ( figure [ fig_fy ] ) . it has a small extension , @xmath106 au , and is globally anti - symmetric . a less evident sub - structure was also present in the previously analyzed mc ( 18 oct . 1995 , ) . if @xmath44 , the central sub - structure would imply the presence of a small twisted flux tube with opposite magnetic helicity in the center of the flux rope . the formation of such a structure is not possible in the corona . after analyzing several possibilities , the simplest interpretation is the following . close to the minimum approach , the spacecraft trajectory is nearly tangent to the magnetic flux surfaces of the flux rope ( @xmath107 ) . any warping of the flux surfaces gives a clear signal in the @xmath41 component . for geometrical reasons , such warping is more difficult to detect outside the center ( where @xmath41 is important ) . it is noteworthy that such structure will be evident only if the magnetic data are rotated to the correct mc frame to have no mixing with the strong @xmath50 component . for the 18 oct . mc , the warping was moderate and @xmath41 kept its sign inside both the in- and the out - bound regions . for the nov . 9 - 10 mc , the warping is more marked as sketched in figure [ fig_center ] . this figure shows ( to the right ) a scheme of the spacecraft trajectory ( vertical dashed line ) across the mc core field lines ( solid warped lines ) . the core crossing starts at field line 1 and ends at 5. the large dot called cc marks the cloud center position . it is located towards positive values of @xmath69 since the cloud axis is above the ecliptic plane ( see last paragraph of section [ mcframe_data ] ) . considering the cloud as a flux rope , which is compatible with the observations , the same accumulated flux @xmath108 implies that the in- and out - bound field lines are connected ( i.e. , they are in fact the same field line observed twice ) . thus , for example 1 is connected to 5. the flux @xmath108 has a local maximum at 2 and 4 and a local minimum at 3 when the spacecraft trajectory is tangent to the field lines . the total azimuthal flux @xmath109 is estimated taking @xmath110 , assuming that @xmath44 . the largest source of uncertainty is the length of the flux rope , @xmath111 , and in a more general way the assumed invariance by translation . since this cloud presents signatures of being detached at one of its legs @xcite , we assume an initial length @xmath112 au , so intermediate between values used in previous papers . the uncertainty in the mc boundary of the out - bound branch has a lower effect since the difference of @xmath109 found with the boundary `` out@xmath98 '' and `` out@xmath97 '' is only about 10% ( see table [ tab_flux ] ) . the axial flux @xmath113 , across a surface perpendicular to the cloud axis , can be estimated directly from the observations assuming a circular mc cross section , @xmath44 and neglecting the expansion to compute @xmath114 . here , we neglect the axial flux in the core since it is a correction of the order of @xmath115 , see . then , for the out - bound branch we compute @xmath116 as ( idem for @xmath117 , but with the integration between @xmath118 and @xmath119 ) : @xmath120 as in previously studied mcs ( e.g. @xcite @xcite , @xcite ; ) @xmath113 is one order of magnitude lower than @xmath109 ( see table [ tab_flux ] ) . the analysis of @xmath94 in section [ direct_method_structure ] indicates that this mc is not formed only by a flux rope . some of the mc characteristics ( see figure [ fig_bv_gse ] ) , such as a very unusually high magnetic field with a low variance , a low @xmath14 , and a strong expansion , continue well after the rear boundary of the flux rope ( `` out@xmath15 '' , `` out@xmath17 '' , and `` out@xmath16 '' ) . we discuss below the most plausible physical scenario to create such magnetic structure . there is an extended region where @xmath41 has still a negative and coherent behavior from position `` out@xmath16 '' to `` back '' , so the accumulated flux keeps increasing monotonously ( figure [ fig_fy ] ) . this behavior was also found in the 18 oct . 1995 mc and it was interpreted as the trace of an original larger magnetic flux rope whose front was partially reconnected with the overtaken magnetic flux , as shown in figure 6 of . in the example analyzed here , this interpretation has even more support from the data , since part of the overtaken flux is still present in front of the mc and we can estimate when reconnection started . there is a coherent negative @xmath41 field just in front of the mc ( from @xmath8 18:00 ut to 20:30 ut , figure [ fig_bv_cloud ] ) . we interpret this as the remnant of the magnetic flux which reconnected with the original flux rope . at the time of observations , the mc overtakes this structure with a velocity difference of @xmath121 km / s ; so reconnection is driven . such velocity difference is crucial for reconnection efficiency since its rate increases with a larger velocity difference @xcite . furthermore , and have found that magnetic holes , such as the one preceding the mc analyzed here ( top panel of figure [ fig_bv_gse ] ) , are associated with magnetic reconnection . taking into account the previous reconnection scenario , the back region , where @xmath122 has still a coherent negative value , is simply the magnetic flux at the periphery of the original flux rope . this back region keeps the properties of typical mcs without fitting in the standard flux rope models . if the front reconnected , this back region is connected to the solar wind field ; so , the magnetic field direction can change because of the propagation of alfven waves , as observed here with the reversal of @xmath123 ( figure [ fig_bv_cloud ] ) . however , if there is no other reconnection process , the flux of @xmath41 can not be removed from the back of the mc ; in fact , we find no other field to allow such reconnection . thus , the closed ( flux balanced ) flux rope observed at 1 au is embedded in a larger structure which includes an extended back region . this flux rope was part of a larger one that was partially pealed at its front because it reconnected with its environment . the high velocity of the mc ( @xmath124 800 km / s ) , compared to its surroundings ( @xmath8 600 km / s ) implies a progressive extension of the back region ( between boundaries `` out@xmath16 '' and `` back '' in figure [ fig_bv_cloud ] ) . assuming that the relative velocity , @xmath125 , was similar at earlier times and that the @xmath41 component ( in the back region ) had initially a value similar to that in the rear of the mc ( because of pressure balance ) , we can estimate the period of time @xmath126 between the start of reconnection and the mc observations . the range of time between the observation of the two extremes of the expanding region ( between `` out@xmath16 '' and `` back '' ) is @xmath127 18 h. this corresponds to a spatial extension of @xmath128 when the region was observed . so , the back region expands in size by @xmath129 since the start of reconnection at the mc front . this extension has its origin in the relative velocity between the mc and the back region ( which becomes lower because of its magnetic connection to the solar wind ) . from the beginning of the reconnection at the mc front , starting a lapse of time @xmath126 earlier than the observations , the `` back '' boundary progressively separates from the mc with a relative velocity @xmath130 . assuming that this relative velocity was not changing drastically during the mc transit , the back region expanded by @xmath131 . equating the previous estimations of the back region extension , we find : @xmath132 the solar event that is the most probable source for the mc studied here occurred in ar 10696 . it is a multiple event that starts with a steep rise in goes light curve at @xmath73 15:50 ut on nov . 7 , and has two clear peaks , one at @xmath73 16:00 ut and the other at @xmath73 16:35 ut . at the time of the second peak a large two - ribbon flare was observed within ar 10696 , but also two h@xmath133 ribbons were seen at both sides of an erupting trans - equatorial filament extending from ar 10696 to ar 10695 at the southwest ( see ) . during this intense event that reached class x2.0 in soft x - rays the full neutral line , which formed a switch - back , erupted . the cme observed in the large angle and spectroscopic coronagraph ( soho / lasco ) c2 on the nov . 7 at 16:54 ut is associated with this multiple flare involving ar 10696 and the erupting trans - equatorial filament to the southwest . have proposed that the source of the mc is the ar eruption , while have considered both possibilities either the ar or the trans - equatorial filament eruption . we discuss below these scenarios in view of our ip analysis . taking @xmath73 15:50 ut on nov . 7 , as the start time of the solar event , and an arrival time for the mc front at @xmath73 20:30 ut on nov . 9 , we obtain a transit time of @xmath73 52 h , comparable to the transit time ( @xmath73 47 h ) computed assuming a constant mc velocity of -800 km / s and the spacecraft located at l1 . both estimations are comparable to @xmath126 ( equation [ t - reconnection ] ) ; this implies that reconnection started when the flux rope was close to the sun . what could be the magnetic structure present in front of the flux rope , probably already from its origin in the corona ? the field in front of the mc is oriented mostly southward with an average field intensity of @xmath73 -30 nt , though there is a structure of northward oriented field with a temporal length between 19:54 ut and 20:01 ut on nov . 9 , an average field strength of @xmath73 -15 nt , and a spatial extension of @xmath73 0.002 au ( taking an average velocity of @xmath73 -800 km / s ) . analyzing the mdi magnetogram on nov . 7 , the most probable origin of the southward oriented structure is the large scale , nearly potential field of ar 10696 . if the source of the mc is the ar eruption ; then , the core of the ar should become kink unstable ( as it was observed in an eruption which occurred in the same ar three days later @xcite ) . in the kink instability , part of the twist is transformed into writhe implying a strong rotation of the flux rope . taking into account the mc orientation found in section [ direct_method_orientation ] , this rotation should be @xmath73 160@xmath33 . the mhd simulation of gives a writhing of the flux tube of @xmath73 120@xmath33 ( before it reconnects with the overlying field ) . numerical simulations by confirm this using a different approach and , indeed , the writhing could be as large as a rotation of 160@xmath33 depending on the properties of the overlying field ( trk , private communication ) . so the amount of rotation and its direction ( for a left - handed flux rope ) , if the mc flux rope comes from the ar , is coherent with recent mhd simulations of kink unstable flux ropes . interplanetary scintillation observations suggest that the material from the core of the ar was ejected primarily northward and , thus , it could remain unobserved at 1 au ( see ) . in this scenario , and if the source of the mc is the trans - equatorial filament eruption , the direction of the trans - equatorial filament and the cloud axis differ by 67@xmath134 ( see figure 12 in ) . moreover , the sense of rotation from the filament to the cloud axis is opposite to the one expected from the development of a kink instability in a region of left - handed magnetic helicity . as discussed by , it may happen that the kink instability has not played a role far away from the strong fields of the ar . the northward oriented structure observed between 19:54 ut and 20:01 ut on nov . 9 could be related to the northward oriented trans - equatorial loops observed above the filament . however , in both scenarios reconnection should be forced between the ejected flux rope and the above ar arcade field from the beginning of the eruption . this explains the origin of a nearly antiparallel field ( neglecting the short period of northward directed field ) in front of the flux rope from the beginning of the launch , as indicated by a value of @xmath126 similar to the transit time . our previous discussion has implications for the reconnection rate in a collisionless plasma . while the erupting flux rope pushed against an overlying nearly anti - parallel magnetic field , we still observe part of this overlying flux in front of the mc at 1 au ! of course we have no way to determine how strong was the forcing , nor the time dependence of the reconnection rate during the flux rope transit . still this observation is a clue that magnetic reconnection is not efficient in a collisionless plasma , as expected from classical theory . some recent observations show direct evidence of magnetic reconnection in a collisionless plasma as the solar wind @xcite . some numerical simulations show that the hall effect can increase the reconnection rate above the classical rate ( e.g. , ) . our observations set a constrain on the reconnection rate , which can be quantified only after a numerical modeling of the flux rope ejection and transit to 1 au . from the computed accumulated flux ( figure [ fig_fy ] ) we can estimate the relative amount of reconnected flux from the original flux rope ejected from the sun as @xmath135% , assuming @xmath101 ( left panel ) . with the expansion length given by equation ( [ length_evol ] ) , the flux present in the back part , which is the fraction of reconnected flux , is @xmath136% ( however it is not obvious that we can use equation ( [ length_evol ] ) for the back part where the field is connected to the solar wind ) . the relative amount of flux present in front of the mc is @xmath137% . this implies that the relative amount of stable ( not kinked ) magnetic flux , arch - like , above the erupting flux rope is in the interval @xmath138$]% of the azimuthal flux in the mc . finally , a fraction of the @xmath41 magnetic flux observed after the boundary `` in '' is expected to be reconnected when the spacecraft exits the flux rope . to estimate an order of magnitude for this flux , let us suppose that the reconnection rate was comparable to its mean value during the transit from the sun . the amount of reconnected flux during the mc observing time is the ratio of the crossing time ( @xmath139 h ) over the transit time from the sun ( @xmath140 h ) times the flux reconnected from the sun . then , @xmath73 5% of the original flux was reconnected during the mc crossing time . so , the fraction of flux reconnected between @xmath141 and @xmath142 is expected to be an amount much lower than the uncertainties in the flux estimations ( see table [ tab_flux ] ) . this cloud presents a large velocity difference between its front and rear parts ( fourth panel in figure [ fig_bv_gse ] ) . this velocity difference is the consequence of a large mc size and a significant expansion . both conditions imply that the magnetic field is observed at significantly different times during the mc crossing , so at times when the mc has significant different sizes . as a consequence of magnetic flux conservation , this implies the observed decay of @xmath143 with time ( upper panel of figure [ fig_bv_gse ] ) . then , @xmath144 , and even more @xmath57 ( since it is stronger close to the flux rope borders ) presents a remarkable asymmetry between the in- and out - bound branches . moreover , the center of the cloud is observed before the central observing time for the full structure , as expected for an spatially symmetric expanding object . in this section we compare the observations to fitted models describing the evolution of the magnetic field assuming an isotropic self - similar expansion . the observed velocity is used to derive the expansion rate , which is then used in the expansion magnetic model . we also quantify the magnetic fluxes from the fitted models , and compare them with the results obtained from a classical static model and the direct method . we assume an isotropic self - similar expansion of the mc , where all distances are multiplied by a factor @xmath145 : @xmath146 where @xmath147 and @xmath148 is the position of a given element of fluid at some reference time @xmath149 . since the mc front boundary , called `` in '' , is well defined , we select this as the reference time , @xmath150 . each element of fluid is labeled by its @xmath148 value , then the @xmath148 coordinate is a marker of each element of fluid ( lagrangian coordinates ) . we assume below that the flux rope size increases linearly with time . then , @xmath145 can be written as : @xmath151 where @xmath152 is the time of expansion . if equation ( [ f_linear ] ) would be valid for all the transit time from the sun , then the flux rope size would vanish at time @xmath153 , so @xmath152 would also be approximately the transit time from the sun to the position observed at @xmath80 . in practice , we use equation ( [ f_linear ] ) only during the mc crossing ; so , we are only assuming a local linear increase of the size with time . depending on the evolution during the transit from the sun , @xmath152 could be different from the transit time . the time evolution for the plasma velocity @xmath154 in the cloud frame comes from equation ( [ r_self_similar ] ) doing the temporal derivative and keeping @xmath148 fixed ( so following a plasma element ) : @xmath155 the first equality gives the velocity at a given time @xmath156 , so it is proportional to the distance . the velocity decays with time , but this decay is only apparent since when one follows a given plasma element , defined by @xmath148 , the velocity is in fact constant ( second equality ) . then , equation ( [ f_linear ] ) implies that there are no forces acting on any plasma element ( free expansion ) . this is an `` a posteriori '' justification of equation ( [ r_self_similar ] ) ; for an isotropic expansion , an initial force - free field stays force - free . the observed speed is the sum of the expansion speed and of the global speed of the mc . during the time of observation , we can assume that the mc is globally moving at a constant speed @xmath157 ( velocity of the mc center ) along @xmath56 because the aerodynamic drag significantly affects @xmath158 only in a time scale comparable or larger than the transit time to 1 au . we also assume that the spacecraft is at rest . then , the velocity component along @xmath56 is : @xmath159 this equation is fitted to the data to derive @xmath158 , @xmath160 ( the time when the spacecraft crosses the cloud center ) , and @xmath152 . it is noteworthy that the nearly linear observed @xmath161 is a consequence of the constant speed @xmath158 , and not of the linear expansion with time assumed in equation ( [ f_linear ] ) . the assumed expansion introduces only a non linear correction in @xmath162 because the crossing time , @xmath163 , of the mc is small compared to the time of expansion , @xmath152 ( a further correction would come from the non - linear development of a general @xmath145 ) . this justifies the use of equation ( [ f_linear ] ) ) . vertical lines mark the same times as in figure [ fig_bv_gse ] . ] in center out@xmath15 out@xmath16 out@xmath17 the observed @xmath164 has a globally linear variation with time within the mc ( figure [ fig_vr_model ] ) . to better show its variation , we subtract the mean value of the velocity computed inside the range of positions `` in '' to `` out@xmath16 '' , @xmath165 km / s . when the impact parameter ( @xmath166 ) is small , as it is in this mc , this speed represents the radial velocity with respect to the cloud axis , such that for a cloud in expansion , @xmath167 is negative before the spacecraft reaches the cloud axis and positive after that . from a least square fitting of equation ( [ v_x ] ) to the velocity data inside the `` in '' to `` out@xmath16 '' boundaries , we obtain @xmath168 h ( @xmath169 days ) . we find a slightly longer time of expansion @xmath170 h ( @xmath171 days ) when the fit is restricted to the interval between the `` in '' and `` out@xmath15 '' boundaries , and we get @xmath172 h ( @xmath173 days ) for boundaries `` in '' to `` out@xmath17 '' . considering the self - similar expansion given by equation ( [ f_linear ] ) , this implies that the mc has expanded from the time @xmath80 by a factor @xmath174 and @xmath175 when the spacecraft crossed `` out@xmath16 '' and `` out@xmath15 '' , respectively . both times of expansion , @xmath176 and @xmath177 , are longer than the transit time from the sun ( @xmath140 h , section [ direct_method_reconnection ] ) . taking an approximately constant global velocity @xmath158 , the distance of the mc front to the sun increases by the factor @xmath178 ( see equation [ length_evol ] ) . doing the ratio of the measured expansion rate variation ( @xmath179 ) to @xmath180 , we obtain an undimensional factor @xmath181 . this factor is 0.64 and 0.59 for boundaries `` out@xmath16 '' and `` out@xmath15 '' , respectively . considering the orientation of the mc ( @xmath54 and @xmath55 ) , the velocity in figure [ fig_vr_model ] is measured mainly across the flux rope . then , the mc is expanding at a significantly smaller rate radially than what we expect assuming the isotropic expansion in equation [ length_evol ] . assuming that the spacecraft trajectory is close enough to the mc axis , @xmath182 , the observed velocity lets us compute the radius of the flux rope since the distance crossed ( @xmath183 ) is @xmath184 . taking the boundary `` out@xmath16 '' , we find @xmath185 au and @xmath186 au ; while for boundary `` out@xmath15 '' we find smaller radii , @xmath187 au and @xmath188 au . if we do not consider the expansion , we get the mean values because the distance crossed is simply @xmath189 . lundquist s static model @xcite is a classical linear force free configuration ( @xmath190 , with @xmath133 constant ) . from the conservation of the magnetic flux during the expansion , and assuming a lundquist s field at a given time , it is possible to derive an expansion field model ( see , e.g. , , ) . we will also consider a more general model where the amplitude of the azimuthal and axial components are independent this represents an approximation for a flux rope with an oblate cross section ( see for an exact solution ) . we consider this modified expansion lundquist s model to keep the same functional dependence of the field components , so the difference between the three models is only the number of free parameters ( and the physics involved ) . the three models are described by the equations : @xmath191 where @xmath192 , @xmath193 , and @xmath194 are the field and @xmath133 values when the spacecraft enters the mc at @xmath141 . for the static model @xmath195 and @xmath196 , so there are only two free parameters , @xmath197 and @xmath194 . for both expansion models @xmath198 is function of time as given by equation ( [ f_linear ] ) . for the expansion lundquist s model , @xmath196 , so there are also two free parameters as for the static model . for the modified model , there is an extra free parameter ; the parameters are : @xmath192 , @xmath193 and @xmath194 . notice that for all models @xmath199 is not forced to vanish at the mc boundaries ( this lets @xmath194 as a free parameter ) . because for a fixed time @xmath156 , each of the components @xmath200 and @xmath199 have a spatial dependence as in lundquist s model , the equations for the magnetic fluxes @xmath113 and @xmath201 are the same as in , but now @xmath202 and @xmath203 with @xmath204 and @xmath111 the radius and the length of the cylinder at time @xmath141 . then , @xmath113 and @xmath109 are , as expected , constants of motion , because the increase in @xmath76 and @xmath205 cancels the decay of the field components and of @xmath133 . they both simply write : @xmath206 ( left panels ) and @xmath144 ( right panels ) . observations ( dots ) , cylindrical dynamical model ( red dotted lines ) , cylindrical static model ( blue dotted lines ) , and modified model ( see text , green dotted lines ) . upper panels show the fitting using the end time as nov . 10 , 10:20 ut ( out@xmath16 ) , lower panels use nov . 10 , 8:15 ut ( out@xmath15 ) as the end time . vertical lines mark the same times as in figure [ fig_bv_gse ] . ] out@xmath16 out@xmath16 out@xmath15 out@xmath15 in this section we fit the mc observations with the three models described in section [ magnetic_models ] . we fix the orientation to the one given in section [ direct_method_orientation ] ( @xmath54 and @xmath55 ) . this allows us to test the effect of including the expansion and the decoupling of azimuthal / axial field from the problem of finding the mc axis . we first use the data in between `` in '' and `` out@xmath16 '' . we use a nonlinear fitting routine to fit the models presented in section [ magnetic_models ] , assuming @xmath44 , to the observations of @xmath57 ( which corresponds to @xmath207 ) and @xmath144 . the static model can not reproduce the observed asymmetry due to the decay of the field and the shift of the position of the cloud center , because of its intrinsic symmetry ( top panels of figure [ fig_b_fitting ] ) . both the static and expansion lundquist s models overestimate the axial field , @xmath144 , near the cloud center and underestimate the azimuthal field , @xmath57 , near the boundaries . when the extra freedom @xmath208 is considered , the model can reproduce significantly better the observations . a quantitative comparison between the different models is given by @xmath209 , where @xmath210 is the time average of @xmath211 . we find @xmath209=14:16:19 nt , for the modified model , expansion and static lundquist s models , respectively . for the three models a shift of the positions where @xmath212 and where @xmath144 is maximum to a time later than in the observations is present ( top panels of figure [ fig_b_fitting ] ) . we also explore the two earlier rear boundaries : `` out@xmath15 '' and `` out@xmath17 '' . a significant better fit is found with the boundary at positions `` out@xmath15 '' and `` out@xmath17 '' ( lower panels in figure [ fig_b_fitting ] show the fitting for `` out@xmath15 '' , very similar fitted curves are obtained for `` out@xmath17 '' ) . a quantitative comparison between the fits is given in the form @xmath213:@xmath214:@xmath215 in units of nt . for the static model we obtain 16:17:19 , for the expansion model 15:15:16 , and for the modified model 12:12:14 . if the boundary is located at `` out@xmath17 '' , the modeled cloud center corresponds now to nov . 10 at 02:16ut , 14 minutes later than the center given by the observations and the direct method in section [ direct_method_structure ] ( dotted line , nov 10 , 02:02ut ) . for a rear boundary at `` out@xmath15 '' , the modeled cloud center corresponds to nov . 10 at 01:58ut , only 4 minutes earlier than the value obtained using the direct method . let us now analyze the differences between the observations and the best model ( modified lundquist s model with boundary `` out@xmath15 '' ) . the model follows globally well the observations , except for the central reversal of @xmath57 and close to the boundaries . this central reversal can not be taken into account by the model ( see section [ direct_method_structure ] for an analysis of this feature ) . close to the boundaries , the asymmetry of the model is not as large as the observed one ; the isotropic expansion model gives @xmath216 , while the observations give @xmath217 . due to the crossing geometry , the data mainly reflect the radial expansion velocity . let us now consider a refinement of the above model , the radial expansion is still given by @xmath145 , while the axial expansion is rather given by @xmath218 ( equation ( [ length_evol ] ) ) . @xmath57 is affected both by the radial and the axial expansion . the conservation of the azimuthal flux gives : @xmath219 $ ] . in the fitting of the models , we have assumed @xmath220 , but in fact @xmath221 so that @xmath222 is slightly larger than @xmath223 . with @xmath224 h and @xmath225 h , we find @xmath226 and @xmath227 , which gives @xmath228 ( rather than @xmath229 with an isotropic expansion ) , a value closer to the observed value ( @xmath217 ) but still larger . a spatial asymmetry between the in- and out - bound branches , which can not be attributed to the expansion , is present in the observations ( figure [ fig_b_fitting ] ) . indeed , the observed @xmath57 does not have the expected decrease towards the front boundary ( which is present in the model and also in the observations towards `` out@xmath15 '' , figure [ fig_b_fitting ] ) . at the mc front the magnetic field is expected to be compressed ( then , it is enhanced ) by the dynamic pressure of the overtaken plasma . the interaction with the surroundings is likely to be at the origin of this extra asymmetry ( on top of the expansion ) between the in- and out - bound branches . . magnetic fluxes present in the flux rope in units of @xmath230 mx using different fitted models or the direct method , for three backwards boundaries ( `` out@xmath15 '' , `` out@xmath17 '' , and `` out@xmath16 '' in figures [ fig_bv_gse ] and [ fig_bv_cloud ] ) , and for orientation given by @xmath54 and @xmath55 . for the direct method we present an average between the values obtained for the in - bound and out - bound branches ( @xmath113 is not affected by including axial expansion ) . the fluxes are not corrected for the flux lost by magnetic reconnection during the mc travel from the sun ( since only the direct method permits an estimation of the reconnected flux ) . the three models are described by the same equations but differ by the constraint set on the free parameters , so the physics involved ( equations [ b_r]-[b_z ] ) . an initial length for the cloud @xmath112 au is assumed . [ cols="^,^,^,^,^ , > , > , > , > " , ] from the fitted parameters and expressions given in equations ( [ fz_ra])- ( [ fphi_ra ] ) , we obtain the values for the magnetic fluxes using the different models ( table [ tab_flux ] ) . we assume an initial length @xmath112 au as in section [ direct_method_flux ] . from table [ tab_flux ] , @xmath113 is estimated in the range [ 2.1 - 7.7]@xmath231 mx , and @xmath109 in the range [ 60 - 102 ] @xmath231 mx . fixing the end boundary at nov 10 , 10:00 ut ( only 20 minutes earlier than boundary `` out@xmath16 '' ) , and using the static classical lundquist s model , reported @xmath232 mx and @xmath233 mx / au ( thus @xmath234 mx , for a cloud length of @xmath235 au , as assumed here ) . these values , as expected , are very close to our present results with the static model and boundary `` out@xmath16 '' . our present results show that the expansion affects slightly the computed fluxes ( more @xmath109 ) , while decoupling the fits of @xmath41 and @xmath123 has the largest effect . for rear boundaries `` out@xmath15 '' and `` out@xmath17 '' , the estimations of both fluxes using the modified lundquist s model are in close agreement with the results of the direct method that consider an axial expansion , in particular this agreement is much better for @xmath109 . considering the rear boundary `` out@xmath16 '' , the value of @xmath236 obtained from the direct method is lower than the one from the modified lundquist s model due to the significantly negative value of @xmath144 beyond `` out@xmath17 '' ( see figure [ fig_bv_cloud ] ) ; this contributes to decrease @xmath236 when integrating the circular cloud section between `` out@xmath17 '' to `` out@xmath16 '' . as discussed in section [ direct_method_orientation ] , we believe that the rear boundary of the cloud should be between `` out@xmath15 '' and `` out@xmath17 '' . the icme of nov 9 - 10 , 2004 , was a complex event with a large expansion and a strong magnetic field in the front decreasing monotonously ( almost linearly ) with time . earlier analysis considered that two magnetic clouds ( mcs ) were located inside this icme . latter studies concluded that only one mc was present . we confirm this and precise the orientation and boundaries of the flux rope using several methods . we also find clues about the interaction of this mc with its surroundings . to facilitate the understanding of the physics involved in the mc it is useful to transform the data to the local mc frame where the axial and azimuthal components of the magnetic field are decoupled . a classical method to determine the mc axis is the minimum variance ( mv ) method , which takes into account the different spatial behavior of the magnetic field components to find the flux rope orientation . we minimize the effect of the strong expansion , which implies a decreasing field magnitude with time , normalizing the field at each data point . this gives a range of possible orientations ( typically with a precision of the order of @xmath237 ) . the determination of the orientation of the flux rope axis is improved using the direct method . this method is based on two main points : first , the flux rope is topologically distinct from the surroundings , so it should generically be bounded by a discontinuity of the magnetic field components ( presence of a current sheet ) , and second , the same amount of azimuthal magnetic flux should be present in the in- and out - bound branches of the cloud . for the cloud of nov . 9 - 10 , 2004 , the frontal discontinuity is well defined , while three rear discontinuities are present , called `` out@xmath15 '' , `` out@xmath17 '' , and `` out@xmath16 '' . they are separated by about 2 h compared to a mc duration of about 14 h. the azimuthal flux relates the frontal discontinuity to two discontinuities at the rear using two different hypotheses for the axial evolution : an expansion comparable to the radial one gives a rear discontinuity at `` out@xmath17 '' , and a negligible axial expansion gives a rear discontinuity at `` out@xmath16 '' . however , since the second case implies a reversal of the axial field at the rear of the mc , we conclude that the first discontinuity ( `` out@xmath17 '' ) is associated to the frontal discontinuity . this defines precisely the extension , as well as the orientation angle @xmath238 ( within @xmath239 , @xmath240 ) of the flux rope . fitting a model to the data is another approach to understand the observed magnetic structure . we have used three models that are based on the lundquist s solution . the first one is the classical static solution . the second one includes a self - similar expansion with the same rate in the axial and radial directions . finally , the third one also includes an isotropic expansion and decouples the fit of the azimuthal and axial field components to take into account the observed stronger azimuthal component ( a possible signature of a flat cross section ) . the expansion rate is obtained fitting the model to the observed plasma velocity . the best fits to the data are obtained when the first and second discontinuities ( labeled `` out@xmath15 '' and `` out@xmath17 '' ) are used , in agreement with the results obtained with the direct method . comparing the results of the fitted models with the direct method , using boundaries `` out@xmath15 '' and `` out@xmath17 '' , we find that the axial and azimuthal fluxes are in the ranges [ 4.7 - 7.7]@xmath231 mx and [ 60 - 95]@xmath231 mx , respectively . the main limitation on the axial flux measurements is the unknown shape of the cross section . for the azimuthal flux , it is important to consider the axial expansion . here the limitations are different , the shape of the cross section is not important , the main limitation is the distribution of the flux along the mc axis . finally , we confirm that the azimuthal flux is one order of magnitude larger than the axial flux . after the large and coherent rotation of @xmath12 , some typical mc characteristics are still present : low level of fluctuations , strong expansion ( observed in the decay of @xmath13 and @xmath143 ) , intensity of magnetic field higher than the typical solar wind values , low @xmath14 . these are evidences that the mc extends farther in the back of the flux rope . part of this back ( from `` out@xmath16 '' to `` back '' in figures [ fig_bv_gse ] and [ fig_bv_cloud ] ) shows a coherent behavior of @xmath57 , which we interpret as the signature of an originally larger flux rope that was partially reconnected in its front near the sun , with the consequent flux removal . reconnection of the cloud field with the overtaken solar wind field is another source of underestimation of the original magnetic flux that was launched from the sun . we have found that @xmath241 % ( with a @xmath242 % as an upper bound ) of the azimuthal flux was lost in the front of the mc during its travel from the sun ; this is much less than in the oct . 18 - 30 , 1995 mc , where it was estimated to be about 57 % @xcite . another difference between the mcs is that the nov . 9 - 10 mc has a back part moving at a speed significantly lower ( by @xmath243 km / s ) than the flux rope ( while the oct . 18 - 20 mc was overtaken by a fast stream ) . then , the reconnected field progressively forms an extended region in the back of the flux rope ( with a weaker magnitude but still with a smooth spatial variation ) . from the extension and velocity difference of this region with the flux rope , we have estimated that reconnection started close to the sun , possibly between the erupting twisted flux tube ( giving the flux rope ) and the overlying arcade of active region 10696 . part of the arcade field is probably present in front of the mc , with a nearly anti - parallel direction and a significant velocity difference ( @xmath244 km / s ) , indicating that magnetic reconnection is not so efficient in the interplanetary space . reconnection in front of the mc has several observational consequences , as follows : + first , it introduces an asymmetry in the observed magnetic field . the remaining part of the flux rope is observed first followed by an extended tail of weaker magnetic field ( which is re - orientated since it has changed its connectivity ) . when this process is dominantly at work , this implies the presence of a closed flux rope at the beginning of the icme . + second , depending on the solar launch direction , the spacecraft could cross the flux rope or its laterally extended back part . in this last case , one would detect some characteristics of the mc ( such as a coherent field and low @xmath245 values ) , but without the coherent rotation of the field . such crossing would be classified as an icme ( without mc ) . an example of such observations with the two helios spacecrafts is analyzed by and another example using the ace spacecraft is analyzed by . + finally , the back flux connected to the solar wind field makes the moving magnetic structure larger in the transverse direction ( orthogonal to the global motion ) . from its mixed origin the back region is expected to move at a speed intermediate between the mc and the solar wind speed , as observed in the present analyzed mc . then , with a significant larger velocity than the surrounding medium , the back region is expected to have an effect on the frontal shock surrounding the icme . a larger transverse scale implies a forward shock at a larger distance in front of the mc than the distance deduced by its flux rope transverse size @xcite . such large distance has so far being interpreted as a flat flux rope @xcite . the consequence of reconnection in the front of the mc is an alternative and/or complementary explanation which requires numerical simulations to be quantified . the results obtained show the potentiality of combining several methods of analysis , minimum variance , direct method and fit to the data . this analysis will be done for other mcs to derive the variety of possible physical scenarios and also to improve our understanding of mcs and icmes . the authors are grateful to l.k . harra for organizing the sun - earth connection workshop at mssl from which this contribution is an outcome . this workshop was possible thanks to a phillip leverhulm prize . s.d . would like to thank n.u . crooker and h. elliott for helpful discussions . this research has made use of nasa s space physics data facility ( spdf ) . c.h.m . and p.d . acknowledge financial support from cnrs ( france ) and conicet ( argentina ) through their cooperative science program ( n@xmath246 20326 ) . this work was partially supported by the argentinean grants : ubacyt x329 , pip 6220 ( conicet ) , and picts 03 - 14163 , 03 - 12187 and 03 - 33370 ( anpcyt ) . s.d . and c.h.m . are members of the carrera del investigador cientfico , conicet . m.s.n . is a fellow of conicet .	the solar wind conditions at one astronomical unit ( au ) can be strongly disturbed by the interplanetary coronal mass ejections ( icmes ) . a subset , called magnetic clouds ( mcs ) , is formed by twisted flux ropes that transport an important amount of magnetic flux and helicity which is released in cmes . at 1 au from the sun , the magnetic structure of mcs is generally modeled neglecting their expansion during the spacecraft crossing . however , in some cases , mcs present a significant expansion . we present here an analysis of the huge and significantly expanding mc observed by the wind spacecraft during 9 and 10 november , 2004 . this mc was embedded in an icme . after determining an approximated orientation for the flux rope using the minimum variance method , we precise the orientation of the cloud axis relating its front and rear magnetic discontinuities using a direct method . this method takes into account the conservation of the azimuthal magnetic flux between the in- and out - bound branches , and is valid for a finite impact parameter ( i.e. , not necessarily a small distance between the spacecraft trajectory and the cloud axis ) . the mc is also studied using dynamic models with isotropic expansion . we have found @xmath0 mx for the axial flux , and @xmath1 mx for the azimuthal flux . moreover , using the direct method , we find that the icme is formed by a flux rope ( mc ) followed by an extended coherent magnetic region . these observations are interpreted considering the existence of a previous larger flux rope , which partially reconnected with its environment in the front . we estimate that the reconnection process started close to the sun . these findings imply that the ejected flux rope is progressively peeled by reconnection and transformed to the observed icme ( with a remnant flux rope in the front part ) .
we consider the dissipative critical dynamics of the quantum phase transitions ( qpt ) taking place in the system of the coupled enharmonic oscillators with one - component order parameter ( @xmath5 ) corresponding , for example , to the ising magnet @xcite . it is well known that at @xmath6 , in the regime of quantum fluctuations ( zero - point fluctuations ) , the ordering phase transition is possible in these systems @xcite . in addition it is believed that the critical exponents of this phase transition are determined with help of the simple rule : the exponents of the phase transition in a @xmath7-dimension system at @xmath6 are the same as at @xmath8 but in the system with greater per unit dimension : @xmath9 @xcite . hence one can conclude that the upper critical space dimension of the considered system is @xmath10 . let us call it quantum mechanical ( qm ) approach . however , when describing the dynamics of the statistical ensemble of the coupled oscillators ( @xmath11 ) one needs to take into account the dissipation effect @xcite . in case of @xmath12 this leads to the change of the critical exponents and the universality class of the phase transition in the one - dimension system : @xmath13 , @xmath14 @xcite . in this paper we describe the critical dynamics of the ising magnet system close to qpt using the keldysh technique @xcite . this approach is developed for description of the dynamics of non - equilibrium quantum systems . therefore , we suppose it will allow us to describe the crossover from the critical dynamics to the quantum critical dynamics ( qcd ) using the uniform technique . we also believe that it help us to outline the borders of applicability of the qm and qcd approaches to the qpt description . let us consider the quantum critical dynamics of the ginsburg - landau model in terms of the keldysh technique . the lagrangian of this model has the following form : @xmath15 where @xmath16 is the scalar order parameter field , which obeys to the bose statistics . we suppose @xmath17 and @xmath18 to depend on some external parameter @xmath19 , that controls the system state . it is convenient to describe the non - equilibrium dynamics of quantum systems in terms of the keldysh technique . since we assume the uniform description of both quantum ( @xmath20 ) and classical ( @xmath21 ) systems , we consider the system interacting with a heat bath at the temperature @xmath22 . according to the keldysh approach to the description of non - equilibrium dynamics of the system one should write the generating functional in the form of @xmath23 where @xmath24 , @xmath25 and @xmath26 are the `` classical '' and `` quantum '' parts of the order parameter accordingly , @xmath27 and @xmath28 are the sources of these fields , and @xmath29 is the fields lagrangian density . below it will be more convenient to move from the minkowski space to the euclidean one by the wick rotation , @xmath30 . then @xmath31 note , that in this case every contact of the system with any environment , including external noise , is described as the interaction with the heat bath , while the `` internal ( quantum ) noise '' is implied in the description directly in the description . in this case according to @xcite one can write the keldysh lagrangian in the form of : @xmath32 where @xmath33 \mathcal{l}_{int}= -\frac{}{}u(\phi_{cl}+\phi_q,\,g_{cl}+g_q)+u(\phi_{cl}-\phi_q,\,g_{cl}-g_q),\\ \mathcal{l}_{noise}=\phi_q \left(2\gamma \omega \coth { \frac { \displaystyle \omega } { \displaystyle t}}\right)\phi_q,\end{gathered}\ ] ] @xmath34 , and @xmath35 is the interaction part . according to the keldysh approach to the description of non - equilibrium dynamics one can write an expression for the retard , advanced and the keldysh parts of the green function ( matrix ) in the form of : @xmath36 where @xmath37 is the hermitian matrix ( @xmath38 ) , and the circular multiplication sign implies integration over the intermediate time ( matrix multiplication ) @xcite . one can check that @xmath39^k=[g^r]^{-1}\circ f - f\circ [ g^a]^{-1 } .\end{gathered}\ ] ] after the wigner transform ( wt ) in the frequency representation we come to @xmath40^k = f(\omega)\left([g^r]^{-1}-[g^a]^{-1}\right),\end{gathered}\ ] ] where @xmath41 is the distribution function . for a boson system in thermal equilibrium @xmath42 , where @xmath22 is the temperature of the heat bath @xcite . this is the fdt , which , as it is shown later , takes a different form in the classical and quantum limits . if we consider the system with dissipation , then @xmath43^{-1}=\varepsilon_k + i\gamma\omega , \quad [ g^a]^{-1}=\varepsilon_k -i\gamma\omega , \\ [ g^{-1}]^k=2\gamma\omega \coth ( \omega / t ) , \label{a3}\end{gathered}\ ] ] where @xmath44 is the kinetic coefficient . in the quantum case @xmath45 ( see fig.[f1 ] ) @xmath46^k=2\gamma \|\omega \|.\end{gathered}\ ] ] the fdt has the following form : @xmath47 . in the classical case @xmath48 ( see fig.[f1 ] ) @xmath49^k=2\gamma t,\end{gathered}\ ] ] and the system satisfies the usual classical form of fdt : @xmath50 . versus @xmath22 function ( with @xmath51 ) , the green line is the @xmath52 function . at high temperatures these graphics coincide , which corresponds to the critical dynamics . however , @xmath53 close to @xmath6 , where the system is described by the quantum critical dynamics . ] below we will concentrate on the quantum limit ( @xmath54 ) , when @xmath55 , the temperature is not essential in the fdt , @xmath56 , and the keldysh green function has the following form : @xmath57 note , that in the case of @xmath58 @xmath59 . this is the so called @xmath60-noise , whose intensity does not depend on the temperature but it is equal to @xmath61 . one can infer that the presence of @xmath60-noise is a natural property of the cold many - body bose system , which follows from the quantum character of dynamics in @xmath6 . we suppose that the system is close to the second order phase transition , when the interaction part of the action can be written as @xmath63 where @xmath64 close to the phase transition point , @xmath2 . below we will consider the quantum limit @xmath20 of the critical dynamics of this system in the @xmath65 space dimension close to the second order critical point . the critical dynamics rests on the hypothesis of dynamic scaling , according to which the action should be invariant with respect to the scale transformations which conformly expand the space and time coordinates ( @xmath66 ) . in this case the summarized dimension , @xmath67 ( @xmath68 is the dynamic exponent ) , has the same role as the conventional ( momentum ) dimension , @xmath69 , in the static case . the canonical dimensions of the fields and the model parameters are determined from the condition of dimensionless action . the corresponding summarized canonical dimensions , @xmath70 $ ] , of any values , @xmath37 , are determined as : @xmath71=d[f]+z\cdot d_{\omega } [ f],\ ] ] where @xmath72 $ ] is the frequency dimension @xcite . the canonical dimensions of the values of our theory are given in the table : [ cols="^,^,^,^,^,^,^,^",options="header " , ] ( a ) , advanced , @xmath73 ( b ) , and retarded , @xmath74 ( c ) , green functions of the theory . ] the renormalization procedure is carried out with the standard method . it is assumed that the fields @xmath75 , and @xmath76 are slow - varying ones , such that the fourier - transformed fields have only long - wave components : @xmath77 ; @xmath78 . at the first step of rg transformations one integrates the partition function over the components of the fields in the limited wave band @xmath79 , @xmath80 . the renormalized parameters have the following form : @xmath81 where @xmath82 and @xmath83 are the constants of renormalization . let us explain the renormalization of @xmath17 as an example in detail . we will limit ourselves to using the one loop approximation , that is quite enough for the demonstration of all the features of the theory . in this case the graphical representation of the main divergent contribution to the renormalization is shown in fig.[f3]b , and the renormalization constant of @xmath84 has the form : @xmath85 \displaystyle = \mu -{\frac { \displaystyle 12\mu v\pi^2}{\displaystyle \gamma ( 2\pi)^3}}\int\limits_{\lambda k_0}^{k_0}{\frac { \displaystyle dk}{\displaystyle k}}=\mu -{\frac { \displaystyle 3\mu v}{\displaystyle 2\gamma \pi}}\ln(1/\lambda ) . \end{array}\ ] ] one can see that the integral in this expression introduces a logarithmically divergent contribution to @xmath17 renormalization if the momentum dimension is @xmath86\equiv d=2 $ ] . in this case we get the following expression for the renormalized value of @xmath17 : @xmath87,\ ] ] where @xmath88 is the logarithmically divergent factor . in the same way one can get other terms of the renormalized action : @xmath89.\ ] ] the contribution to the renormalization of the kinetic coefficient @xmath90 is proportional to @xmath91 : @xmath92 hence , in the one - loop approximation the renormalization group of the model under study has the form : @xmath93 \displaystyle { \frac { \displaystyle \partial \ln v}{\displaystyle \partial \xi}}= \varepsilon -{\frac { \displaystyle 9}{\displaystyle 2}}{\frac { \displaystyle v}{\displaystyle \gamma \pi}}. \end{array}\ ] ] from the condition of the stable point existence , @xmath94 , we obtain @xmath95 . note that in the case of @xmath96 ( @xmath97 ) we get @xmath98 . in this case only the quadratic term is relevant so that with @xmath99 the critical behavior is well described by the gaussian theory . from the above one can conclude , that there is the quantum order - disorder phase transition in one - dimensionalal systems . this differs appreciably from the classical case , in which the thermal fluctuations control the relaxation processes . however , this result agrees with the experimental data for the quasi - one - dimensional systems . also , from ( [ rg1 ] ) one can see that in the one loop estimation the critical exponent @xmath100 , and one can predict that the relaxation time , or equivalently , the phase coherence time , diverges at the critical point @xmath2 as @xmath101 for bose systems we have formulated the crossover from the critical dynamics regime of qpt to the quantum critical dynamics regime in the framework of the non - equilibrium quantum fields theory in the keldysh technique . the key point of this crossover is that the random noise becomes pink when quantum fluctuations dominate over thermal fluctuations , @xmath102 . as a result , the system goes into a different class of universality , and the critical interval shifts into the low dimension area : @xmath103 . experimental observations of qpt show that they can take place in one - dimensional systems . this conforms both with the qm approach and with the qcd description . however , according to qcd in the case of two - dimensional system at @xmath104 the critical exponents should approach the mean - field theory values . this conforms with the recent experimental results @xcite , and distinguishes this theory from qm approach , in which it should be possible only in the case of @xmath105 . one can assume that the reason of this is following : indeed , at the time scales which are much more then the phase coherence time the description of the phase transition needs the qm approach . however , at the phase transition the phase coherence time diverges , and the experimental time scales can not exceed it . as a result the observed critical behaviour corresponds to the qcd description . subir sachdev , _ quantum phase transitions _ ( cambridge university press , isbn 0521004543 ) , 2001 p. 353 ; m.i . kaganov , a.v . chubukov , sov . 30 10151040 ( 1987 ) ; s. pankov et al . , b * 69 * 054426 - 054436 ( 2004 ) ; u. weiss , quantum dissipative systems . world scientific , singapore , ( 1999 ) ; c. hohenberg and b.i . halperin , rev . phys . * 49 * , 435 ( 1977 ) ; a. kamenev and a. levchenko , advances in physics 58 , 197 ( 2009 ) ; a.n . vasilev , _ quantum - field renormalization group in the theory of critical phenomena and stochastic dynamics _ ( crc press , boca raton , london , new york , washington , isbn : 0415310024 ) , 2004 , p 705 ; a.z . patashinskii , v.l . pokrovskii , 1979 _ fluctuation theory of phase transitions _ ( pergamon press , oxford , new york , toronto , sydney , paris , frankfurt ) p 321 ; x. zhang , c.l . hung , s.k . tung and c. chin , science 335 , 1070(2012 ) ;	the quantum critical dynamics of the quantum phase transitions is considered . in the framework of the unified theory , based on the keldysh technique , we consider the crossover from the classical to the quantum description of the boson many - particle system dynamics close to the second order quantum phase transition . it is shown that in this case the upper critical space dimension of this model is @xmath0 , therefore the quantum critical dynamics approach is useful in case of @xmath1 . in the one - dimension system the phase coherence time does diverge at the quantum critical point , @xmath2 , and has the form of @xmath3 , the correlation radius diverges as @xmath4 .
the world airline network ( wan ) has massively increased the speed and scope of human mobility . this boon for humanity has also created an efficient global transport network for infectious disease @xcite . pandemics can now occur more easily and more quickly than ever before . the accelerating emergence of novel pathogens exacerbates the situation @xcite . better understanding of global dispersal dynamics is a major challenge of our century @xcite . rapid assessment of an emerging outbreak s dissemination potential is critical to response planning @xcite . we do not know where the next pandemic threat might emerge . mexico was not a prime candidate for an influenza outbreak , nor west africa for ebola . preemptively mapping the pandemic influence of individual airports could contribute substantially to monitoring and response plans . while exact relationships between the wan and pandemic spread are difficult to model @xcite , simulation studies suggests that topological descriptors which describe epidemic outcomes on network models also have explanatory power for relationships between the topology of the wan and pandemic spread @xcite . observational studies of influenza @xcite , malaria @xcite , and dengue fever @xcite support this conclusion . given the topology of a network , the minimal disease transmission rate which allows epidemics is given by the inverse of the spectral radius of a network s adjacency matrix @xcite , and the typical outcome @xcite and time course @xcite of an epidemic follow a closed - form solution governed by the degree distribution of the network . the wan s topological structure is well characterized . it is a small - world , scale - free network with strong community structure , imposed partly by spatial constraints @xcite . the majority of airports ( 70% ) serve as bridges which connect a densely interconnected core of 73 major transport hubs ( 2% ) to regional population centers and peripheral airports ( 28% ) @xcite . nodes which connect communities can be distinct from high - degree nodes within communities @xcite . since the wan is designed to optimize passenger flow , the network s temporal structure has little effect at time scales relevant for pandemic spread @xcite . topological descriptors of epidemic dynamics , however , can only describe typical outcomes . they do not describe the structure of the variation around the typical outcome , which is dismissed as stochastic when mentioned at all . even within the constraints of a simple branching process model , empirical estimates of the probability of epidemic show substantial variation around the analytically derived solution , see figure [ fig : branching ] . actual outcomes of emergent infectious diseases are crucially shaped by chance events in the early phases of their emergence @xcite . clear understanding of how seed location influences global outcomes would substantially improve public health planning @xcite . the development of sophisticated , parameter - rich epidemic simulators provides powerful tools for exploring relationships between seed location and epidemic outcomes @xcite . common frameworks encompass demographic and mobility characteristics via either metapopulation @xcite or agent - based assumptions @xcite . careful tuning of these models has produced results which well match the spread of the 2009 influenza epidemic @xcite . yet the complex interactions between model structure , input parameters , and estimation methods makes interpretation of model - based results challenging @xcite , especially when attempting to generalized to future outbreaks for which epidemic parameters are fundamentally unknowable . if , however , two radically different modeling approaches result in such high agreement both with each other and with reality @xcite then the principal driver of outcomes should be expressible with a small parameter set @xcite . evidence suggests that simple probabilistic models incorporating local incidence , travel rates , and basic transmission parameters are sufficient to predict outcomes of complex metapopulation based simulations @xcite . recent theoretical work suggests that the apparent stochasticity in the early phases of a network - mediated epidemic process can be explained by the expectation of the force of infection of epidemic processes seeded from that node @xcite . the aim of this study is to evaluate if this finding generalizes to realistic scenarios of wan - mediated pandemic disease spread . our model of the wan is based on the 2014 release of the open flights database @xcite . we selected all airports serviced by regularly scheduled commercial flights , resulting in a list of 3,458 airports connected by 68,820 routes served by 171 different aircraft types . we simplify the network by replacing multiple routes between airports by a single edge whose weight is the sum of the available seats on all routes connecting the two airports , under the assumption that the aircraft type reflects the airline s best judgment of the importance of the route . aircraft seating capacity was estimated based on aircraft descriptions on worldtrading.net and airliners.net , using airlinecodes.co.uk to translate the iata aircraft codes into aircraft type . the expected force of a network node is defined as the expectation of the force of infection generated by an epidemic process seeded from the node into an otherwise fully susceptible network , after two transmission events and no recovery @xcite . the force of infection in a network is directly proportionate to the current number of infected - susceptible edges or in a weighted network , the sum of all such edge weights . its expectation after two transmission is given by the entropy of the distribution of this sum over all possible ways the first two transmission could occur . applied to the wan , @xmath0 where @xmath1 is * * a*irport i s * e*xpected * f*orce , @xmath2 enumerates all possible ways to observe two transmissions seeded from @xmath3 , @xmath4 is the weighted degree of the @xmath5 transmission pattern multiplied by the probability that this pattern is observed given @xmath2 , and @xmath6 is the normalization of @xmath4 . we here further normalize aef values to the range @xmath7 $ ] . epidemic outcomes are generated using the gleamvis simulator @xcite . gleamvis integrates real - world global population and mobility data with an individual based stochastic mathematical model of the infection dynamics to produce realistic simulations of the global spread of infectious diseases . our basic experimental setup is to simulate the same disease model over a range of seed cities . the structure and parameters of the disease models are based on those which match the 2009 influenza pandemic as reported in @xcite and validated in @xcite , specifically , a * susceptible-exposed-infected-r*ecoverd ( seir ) model with transmission rate @xmath8 specified below , latency rate @xmath9 and recovery rate @xmath10 . rates are expressed in units of days . the initial population distribution is 10% of the seed city infected and the remainder of the ( world ) population susceptible . seasonality effects are not included , since their influence varies both by time and geographic latitude , masking variability attributable to seed location . gleamvis divides the world into sixteen regions . an outbreak is declared a pandemic on the day prevalence in at least three regions is greater than one per 100,000 inhabitants . the pattern the results is invariant to thresholds in the range @xmath11 $ ] per 100,000 inhabitants and to replacing the `` three regions '' criteria with `` 100 cities . '' results for each airport are reported in terms of the median over 20 runs ( the maximum number supported by the public gleamvis client ) . if the threshold is not passed after 365 days ( the maximum length supported by the public gleamvis client ) , we declare that no pandemic occurred . for an outbreak to become a pandemic , its basic reproductive number @xmath12 must surpass the basic epidemic threshold @xmath13 needed to establish a disease in a local population by a sufficient amount to also overcome finite subpopulation size effects and diffusion rates to neighboring populations . a branching process approximation suggests that invasion thresholds in metapopulation models depend on the outbreak s @xmath12 value , the variance of the network s degree distribution , and the mobility rate between subpopulations @xcite . the gleamviz model specifies the last two values , reducing invasion thresholds to a function of @xmath12 . however , as shown in figure 1 , even a pure branching process shows substantial variability around the theoretical probability of achieving a large outbreak . for pandemics mediated by the wan , the question of interest is how the invasion threshold varies for different airports . we empirically observe invasion thresholds on the wan as follows . ten seed airports are selected , one from each decile of the range of aef values , see table [ tab : seeds ] . the basic reproductive number is defined as @xmath14 , the transmission to recovery ratio . keeping @xmath15 fixed , we vary @xmath8 over the range [ 0.4 , 0.5 ] , and observe which seeds trigger a pandemic at each value under the simulation framework described above . for @xmath16 , no simulations reached pandemic status , and for @xmath17 , all simulations resulted in a pandemic . often , diseases of concern are known to be competent of invading the network . here , the outcome of interest is not if a pandemic occurs , but rather how long until an outbreak reaches pandemic status . we measure relationships between aef and time to pandemic status as follows . one hundred world airports were chosen such that they evenly cover the range of measured aef values . the simple seir model used previously is extended to the full model used by the gleamvis group to replicate the 2009 pandemic @xcite . this estimates transmission rate as @xmath18 , well above the invasion threshold of @xmath19 determined empirically above . further , the infected compartment of the seir model is divided into three categories : asymptomatic , symptomatic travelers , symptomatic non - travelers . these categories affect the mobility model , and non - symptomatic individuals have reduced transmissibility . for each seed location , we observe both the number of days until pandemic status is reached and the number of days until peak global incidence . both outcomes are highly correlated , since once pandemic status is achieved further disease development is determined by network topology . the purpose of measuring peak global incidence is that this measure is unambiguous , while any definition of `` first day of pandemic status '' is somewhat arbitrary . a shapiro - wilks test of the observed times to peak global incidence suggests that this data is approximately normally distributed ( @xmath20 under the null hypothesis that the data is normally distributed ) , while the distribution of observations of first day of pandemic status is right - skewed ( @xmath21 ) . relationships between outcomes and aef are measured by pearson correlation . we additionally test correlations to weighted and unweighted versions of each airport s betweenness , degree , and eigenvalue centralities , and also to verma et al s t - core , a variant of the k - core which counts triangles @xcite . the robustness of aef values is examined by observing their relative change while progressively degrading the model wan from which they are derived . the network is degraded by removing from one to 15 percent of u.s . airports from the network along with their associated edges . community - based analysis of the wan suggest that us airports form one large community @xcite . the aef of all remaining world airports is computed . three different random removal schemes were used : uniform over all airports , selection weighted by airport degree , selection weighted by aef . the resulting aef values are compared with the original aef values . we record the number of airports whose degraded aef departs from its original aef by more than 1% and by more than 5% . reported results are the average over ten runs , and show the amount of degradation for both u.s . and non - u.s . the aef of the seed location is strongly predictive of an outbreak s invasive threshold as shown in figure [ fig : fitbeta ] and table [ tab : seeds ] . the correlation between aef and the minimal observed transmission rate at which it first became pandemically competent was 0.90 ( 95% confidence interval : 0.98,0.62 ) . tokyo was a notable outlier , achieving pandemic competence earlier than predicted from its aef value . aef was also strongly correlated with the delay until an outbreak became a pandemic . correlation was @xmath22 to the day pandemic status was achieved , and @xmath23 to the day of peak global incidence , see figure [ fig : ttp ] . aef is significantly and more strongly correlated to either epidemic outcome than any of the comparison network centrality measures , see table [ tab : cor ] and figure [ fig : mcomp ] . the aef proved robust to incomplete sampling . degradation was most severe when airports were preferentially removed based on degree . still , only three percent of non - u.s . airports showed more than 1% change in their computed exf values when applying this scheme at the highest noise level . even within the united states , only 22% of aef values changed by more than 5% . see figure [ fig : deg ] . in all cases , aef explains much of the variation in epidemic outcomes , suggesting that the early development of a pandemic is not stochastic , but rather strongly structured by the local connectivity of the seed location . the ability of the aef to summarize this connectivity contributes substantially to our understanding of the role of individual airports in pandemic diffusion . these results are in harmony with other recent work claiming that relative arrival times of wan - mediated pandemics are independent of disease - specific parameters @xcite and that a simple branching process model describes early developments as well as complex metapopulation simulations @xcite . degradation of the network had , in general , limited effect on airport aef values . wrong information regarding a specific node could , however , produce a misleading aef value for that airport . epidemics seeded from airport pbj ( paama island , vanuatu ) took longer than expected to achieve pandemic status . this airport is probably mischaracterized in the open flights database , as flights to this simple grass strip are not shown on the vanuatu airlines online booking system ( ` http://www.airvanuatu.com/ ` , last visited 23 march 2015 ) . in the opposite direction , narita airport ( nrt , tokyo , japan ) showed significantly greater pandemic risk than predicted by its aef . this could be due to japan s intense population density combined with high local mobility , factors captured in the gleamvis simulator but not the open flights database . two outliers highlight a structural blind spot of the aef metric . epidemics seeded from airports zrj ( round lake , canada ) and pvh ( porto velho , brazil ) took longer than expected to achieve pandemic status . zrj is part of a small but locally dense community of airports serving first nation communities in canada . this community has limited connectivity to the rest of the wan , and zrj is three flights distant from any airport outside this community ( winnipeg s james armstrong richardson airport ywg , chicago midway mdw , toronto pearson yyz ) . likewise , pvh is two flights from any of brazil s international transport hubs . the aef is here derived from an airport s two - hop neighborhood , meaning for certain airports it is unaware of these network community boundaries . this limitation could perhaps be overcome by instead computing aef based on a three - hop neighborhood . given , however , that the wan s effective diameter is four hops , and the general good performance of the aef , it is not clear that such an extension would substantially improve results globally . airport expected force summarizes the size , density , and diversity of each airport s neighborhood in the wan . it combines features of degree , neighbor degree , and betweenness centrality in a statistically coherent manner . airport degree is not a good descriptor of pandemic outcomes , since it does not account for a neighbor s onward connectivity . guimera et al noted that high degree does not well correlate to high centrality @xcite . nor does low degree correlate to an airport s connection to the wider network , as illustrated by comparing sweden s linkping city airport ( lpi ) to alaska s huslia airport ( hsl ) . hsl has four outbound routes which connect to other rural alaskan airports . lpi has only one outbound route , which connects to amsterdam schipol . verma et al propose instead characterizing airports based on the number of network triangles they take part in , the t - core @xcite . plotting airport t - core against epidemic outcomes shows that its ability to explain epidemic outcomes is a result of its ability to successfully segment the wan into core and periphery , see figure [ fig : mcomp ] . thus t - core and aef capture complementary aspects of an airport s role in the wan . the applicability of the aef could be extended by modifying it to allow for varying transmissibility at individual airports . such an extension would allow it to express differences in i.e. competent vector species populations or health care system readiness at different world locations . since the aef is the expectation of the force of infection , such an extension merely requires modifying the calculation of each transmission pattern s force of infection along with the probability of that specific pattern occurring . both criteria can be met by adjusting edge weights in the underlying network model , implying that this extension could be implemented using the same framework as outlined in the current work . it would also be interesting to apply the expected force framework to disease spread through the world shipping network , a major transport system for several vector born pathogens along with their vector species @xcite . the approach could also be tested on more local transmission network models , such as contacts in a hospital ward @xcite or city - wide mobility data acquired from i.e. mobile phones @xcite . an outbreak s debut location is highly influential in its ability to become a pandemic threat . the aef metric succinctly captures this influence , and can help inform monitoring and response strategies . these investigations pave the way for the development of simple , robust models capable of informing preparedness planning and policy directives . the max planck society has filed for a patent on the use of the expected force metric to assess spreading risk on the world airline network . gl concieved and carried out the experiments and wrote the manuscript . we thank the gleamvis team for providing public access to their simulator with the only requirement being appropriate citation . trivik verma provided measures of airport t - core . 28 # 1isbn # 1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1#1prebibitemshook , , : , ( ) . doi : : ( ) , ( ) . doi : , , , , , , : ( ) , ( ) . doi : , : ( ) , doi : , , , , : ( ) , ( ) . doi : , , , : ( ) , ( ) . doi : , , : . , ( ) , , , , , , , , , , , : , ( ) . doi : , : , ( ) . doi : , , , , , , , , : ( ) , ( ) . doi : , , : . ( ) , ( ) . doi : : ( ) , ( ) : . ( ) , ( ) . doi : , , : . , ( ) , , : , ( ) . doi : , , , : ( ) , ( ) . doi : , : path lengths , correlations , and centrality in temporal networks . e 84*(016105 ) ( 2011 ) , , , : ( ) , ( ) , , , , , , , , : , ( ) . doi : , , , , , : , ( ) . doi : , , , : ( ) , ( ) . doi : , , , , , , , : , ( ) . doi : , , , , , : ( ) , ( ) . doi : : , ( ) . doi : : openflights . openflights.org , , , , , : , ( ) . doi : , , , , , , , , : ( ) , ( ) . doi : , , , , , , , : ( ) , ( ) . , shown above as the solid blue line , where @xmath12 is the base reproductive number of the disease process . the black dots show empirically observed probabilities from simulations of the same model . each dot is the observed fraction of major outbreaks out of 100 simulated outbreaks for a given value of @xmath12 . for each value of @xmath12 , 100 dots are generated.,scaledwidth=90.0% ] .,scaledwidth=90.0% ] a[fig : deg ] .seed locations . the following airports were selected as seed locations for testing relationships between aef and invasion risk . the table additionally reports the number of days for an outbreak to reach pandemic status ( `` pand . '' ) at the minimal observed transmission rate ( @xmath8 ) for which a pandemic occured , along with each airport s t - core , ( un)weighted degree , and ( un)weighted eigenvalue centralities . [ cols=">,<,<,<,>,>,>,>,>,>,>,>",options="header " , ]	massive growth in human mobility has dramatically increased the risk and rate of pandemic spread . macro - level descriptors of the topology of the world airline network ( wan ) explains middle and late stage dynamics of pandemic spread mediated by this network , but necessarily regard early stage variation as stochastic . we propose that much of early stage variation can be explained by appropriately characterizing the local topology surrounding the debut location of an outbreak . based on a model of the wan derived from public data , we measure for each airport the expected force of infection ( aef ) which a pandemic originating at that airport would generate . we observe , for a subset of world airports , the minimum transmission rate at which a disease becomes pandemically competent at each airport . we also observe , for a larger subset , the time until a pandemically competent outbreak achieves pandemic status given its debut location . observations are generated using a highly sophisticated metapopulation reaction - diffusion simulator under a disease model known to well replicate the 2009 influenza pandemic . the robustness of the aef measure to model misspecification is examined by degrading the underlying model wan . aef powerfully explains pandemic risk , showing correlation of 0.90 to the transmission level needed to give a disease pandemic competence , and correlation of 0.85 to the delay until an outbreak becomes a pandemic . the aef is robust to model misspecification . for 97% of airports , removing 15% of airports from the model changes their aef metric by less than 1% . appropriately summarizing the size , shape , and diversity of an airport s local neighborhood in the wan accurately explains much of the macro - level stochasticity in pandemic outcomes .
the feynman propagator @xmath0 can be interpreted as the time evolution of a wave function @xmath1 that is initially localized at @xmath2 , @xmath3 . in the semiclassical limit @xmath4 can be obtained from the classical trajectories connecting the initial point @xmath2 to the final point @xmath5 in time @xmath6 . the formula which does this is known as the van vleck approximation . one is more likely , however , to be interested in the propagation of a smooth wavepacket , say @xmath7 , rather than an eigenfunction of the position or momentum operators . the straightforward way to accomplish this propagation , given the knowledge of @xmath8 , is with an extra integration , thus @xmath9 in the semiclassical limit this formula becomes @xmath10 one problem with this approach is that there are usually several classical trajectories going from @xmath2 to @xmath5 and they are hard to find , the so - called root - search difficulty . we shall return to a discussion of this point in section 7 . in the present paper , we shall accept the root - search problem and start from equation ( [ ba1x ] ) . the main purpose of this paper is to derive various approximations for eq.([ba1x ] ) and to test them in simple problems . hence we restrict ourselves here to systems with one degree of freedom . in principle , all the expressions that we obtain are easily generalized to multi - dimensional systems . in practice , of course , the difficulty of the numerical calculations increases very fast with the dimensionality , and it also depends on the actual approximation used . we shall evaluate the integral by the method of steepest descent which is also often called , somewhat inaccurately , stationary phase . we shall see that the stationary point is generally complex , giving rise to complex classical dynamics . such a calculation was performed first by heller and collaborators @xcite . the result that they derived can be shown to be identical to our eq.([ba14 ] ) . they used it to calculate the motion of a wavepacket in the morse potential . their work does not seem to have received from physicists all the recognition that it deserves . perhaps this was because there were two rather lengthy papers which contained many somewhat unfamiliar notions . their first paper @xcite reached the result through heuristic arguments and the second one @xcite , though more direct , made many references to the first . in our paper , on the other hand , we shall present in section 2 a very simple and very short derivation of heller s formula , eq.([ba14 ] ) . this basic complex approximation will be the starting point for all other developments of this paper . we shall show that it can be handled for relatively simple problems , but we shall also derive three subsequent approximations involving only real trajectories , which are different for each of the three cases . one case yields the well - known heller thawed gaussian approximation ( tga ) @xcite . the other two give different results and we shall show that they can be quite accurate in some situations . we shall illustrate the application of the various semiclassical formulae with two very simple examples ( section 4 ) and two not so simple ones ( sections 5 and 6 ) . we hope that our derivation of the basic formula and our examples comparing the several approximations will stimulate new interest in semiclassical formulae derived from complex trajectories . in order to simplify our calculations we take for initial state @xmath11 a gaussian wavefunction with average position and momentum @xmath12 and @xmath13 respectively , and position uncertainty @xmath14 . this state is also the coherent state @xmath15 of a harmonic oscillator of mass @xmath16 and frequency @xmath17 . it is defined by @xmath18 with @xmath19 the harmonic oscillator ground state and @xmath20 in the above @xmath21 , @xmath22 , and @xmath23 are operators ; @xmath12 and @xmath13 are real numbers ; @xmath24 is complex . the parameters @xmath25 define the length and momentum scales , respectively , and their product is @xmath26 . the final form of eq.([ba1x ] ) , our semiclassical approximation for wavepacket propagation , is then @xmath27 before we perform the integral in equation ( [ ba3a ] ) by stationary phase , we recall the relations between the elements of the tangent matrix @xmath28 and the second derivatives of the action function @xmath29 computed along the real classical trajectory connecting @xmath2 to @xmath5 in time @xmath6 . given a classical trajectory @xmath30 , @xmath31 with @xmath32 and @xmath33 , its tangent matrix @xmath28 connects , in the linearized approximation , a small initial displacement @xmath34 about the trajectory at @xmath35 to the propagated displacements @xmath36 at time @xmath6 . the relation between @xmath28 and the second derivatives of the action is @xmath37 where @xmath38 , @xmath39 and @xmath40 . a derivation of this formula can be found , for instance , in sec.(2.6 ) of @xcite . notice that we define @xmath28 using the coherent state scales @xmath41 and @xmath42 . inverting this equation we obtain @xmath43 we shall write our final results in terms of elements of the tangent matrix @xmath28 , rather than in terms of derivatives of the action . in this notation the van vleck propagator is @xmath44 for short times @xmath45 is positive and the square root is well defined . for longer times @xmath45 may become negative by going through zero . this happens at the focal points . at these points the van vleck formula diverges and must be replaced by a higher order approximation involving airy functions . sufficiently away from the foci the approximation becomes good again , as long as one replaces @xmath45 by its modulus and subtracts a phase @xmath46 for every focus encountered along the trajectory . we shall not write these so - called morse phases explicitly because they will cancel out in our final formula . the overlap @xmath47 is given by @xmath48 in the semiclassical limit the propagated wavepacket ( [ ba3a ] ) is then @xmath49 with @xmath50 - \frac{(x_i - q)^2}{2 b^2 } -i\pi/4 \;.\end{aligned}\ ] ] we shall now approximate the integration over @xmath2 by the stationary phase method . to do so we assume that the pre - factor in ( [ ba7 ] ) is a slowly - varying function of @xmath2 , which means that the stationary point @xmath51 will be computed by imposing zero variation on @xmath52 alone . the pre - factor will be simply computed at @xmath51 and will not be expanded . we refer to section 3.3 of @xcite for a discussion of this procedure . we need to calculate the first and second derivatives of @xmath52 . the first derivative is @xmath53 where we have defined @xmath54 the second derivative is @xmath55 the stationary position @xmath56 , solution of the stationary phase condition @xmath57 , is given by the equation @xmath58 where we have used @xmath59 and we have made the definition @xmath60 equation ( [ ba9x ] ) makes it clear that @xmath51 is usually complex , and therefore the stationary classical trajectory itself is complex . it leaves @xmath51 at time @xmath61 with momentum @xmath62 , also complex , and it arrives at @xmath5 , which is real , at time @xmath6 . it goes without saying that , for such a complex trajectory , the tangent matrix @xmath28 is complex as well . to integrate over @xmath2 , we expand @xmath52 around @xmath51 to second order , which means that we write @xmath63 . if im@xmath64 is less than @xmath65 at @xmath51 , then re@xmath66 has a negative definite quadratic term and the integral can be performed in the gaussian approximation . assuming this to be the case we find @xmath67 finally , using equations ( [ ba10 ] ) , ( [ ba11 ] ) , ( [ ba4a ] ) , we notice that @xmath68 this simplifies the pre - factor in ( [ ba12 ] ) and we get @xmath69 } \;.\end{gathered}\ ] ] we have changed the subscript on @xmath70 from `` sc '' to `` ct '' to indicate that this semiclassical approximation is calculated with complex trajectories . equation ( [ ba14 ] ) , the semiclassical limit of the propagated wavepacket , is the heller result @xcite . it is the basic result from which all other approximations in this paper are derived . it is _ not _ an initial value representation and it may involve more than one complex classical trajectory . at first glance one might think that this formula represents a frozen gaussian , since the quadratic term in the exponent has a fixed width @xmath41 . a closer look , however , reveals that , since the classical trajectory is complex , all other terms in the exponent also contribute a real part , which changes the effective width : the complex character of the stationary trajectory makes the wavepacket thaw and , at the same time , assume non - gaussian shapes . notice that no morse index or phases of @xmath71 appear in ( [ ba14 ] ) . at @xmath72 one has @xmath73 , @xmath74 one chooses the positive square root and the overlap ( [ ba6 ] ) is recovered . for other times one simply follows the phase of the complex number @xmath75 to get the phase of the pre - factor . before we close this section we re - write the boundary conditions for the complex trajectories in a more convenient form . the stationary trajectory starts at @xmath51 ( usually complex ) and ends at @xmath5 ( always real ) . the initial momentum ( also complex ) is @xmath62 , given by ( [ ba11x ] ) and ( [ ba11 ] ) . since the coordinates @xmath30 and @xmath31 along the trajectory are both going to be complex , it is convenient to define the variables @xmath76 where @xmath77 in general . in this notation hamilton s equations read @xmath78 finally the stationary condition ( [ ba9x ] ) can be re - written as @xmath79 which is equivalent to @xmath80 . therefore the stationary trajectories are solutions of hamilton s equations satisfying @xmath81 the stationary phase approximation of the previous section replaces the integral over a continuum of classical real trajectories by a few complex ones . finding complex trajectories , however , is generally harder than finding real trajectories and one is tempted to look for further approximations to eq.([ba14 ] ) in terms of real trajectories only . these approximations should be good if the stationary complex trajectory is sufficiently close to a real trajectory . what real trajectory ? according to ( [ bcf ] ) , the complex trajectory is determined by two pieces of data : a final position @xmath5 which is real and an initial coordinate @xmath82 which is complex ; see ( [ bc ] ) . the latter is made up of a real position @xmath12 and a real momentum @xmath13 . thus we have a total of three real parameters , @xmath5 , @xmath83 and @xmath84 that could be used to determine a real trajectory . since it actually takes only two parameters , we have three obvious ways of choosing a real trajectory that might sometimes be a good approximation to the complex one . the first way is to give the two initial conditions @xmath85 , which are the center of the initial wavepacket in phase space . the other two ways use the final position @xmath5 with a single one of the initial conditions , either @xmath86 or @xmath87 . each of these three choices leads to a possible approximation in terms of a real trajectory and , as we shall see , they are all different . in this section we shall derive formulae for each of these three cases . the extent of their validity will be examined in later sections . we can try to get a rough sense of how close these three real trajectories are to the unique complex one . the latter goes through @xmath88 hence the @xmath86 and @xmath87 trajectories have at least the merit of being at the right position for the final time @xmath6 . to judge the closeness at time 0 , we can look at formula ( [ ba14 ] ) which tells us that , if im@xmath89 is sufficiently small compared to @xmath41 , the semiclassical wave function at time @xmath6 will be very small unless re@xmath90 is of order @xmath41 or smaller . this is a small distance in the semiclassical limit since @xmath41 is of order @xmath91 . we can also estimate the closeness of the initial momenta by looking at the stationary phase requirement ( [ ba9x ] ) , separating its real and imaginary parts thus @xmath92 again assuming im@xmath89 sufficiently small compared to @xmath41 , we see from the second equation that re@xmath93 is of order @xmath42 or smaller , and @xmath42 is again proportional to @xmath91 . the conclusion is that a semiclassical approximation in terms of real classical trajectories might be useful if conditions are right . to begin , we choose the real trajectory defined by the two initial conditions @xmath85 , the central trajectory of the packet . we call @xmath94 the real end point of this trajectory at time @xmath6 . this trajectory will now be the 0-order approximation . all values of @xmath95 and its derivatives that occur in the equations , including the @xmath28 matrix , will be taken for this trajectory , unless we state otherwise explicitly . all these quantities are now real . we write @xmath96 the stationary phase condition ( [ ba9x ] ) , can be re - written as @xmath97 . then equation ( [ ba15c ] ) can be solved for @xmath98 @xmath99 to simplify the notation , the complex number @xmath100 will be called @xmath101 . we proceed to expand the exponent in ( [ ba14 ] ) about the real trajectory . as before , the pre - factor is assumed to be a slowly - varying function and will be simply computed at the real trajectory . we write @xmath102 the linear terms in @xmath98 cancel . the quadratic terms can all be written in terms of @xmath103 using ( [ ba16 ] ) . all second derivatives of the action can be written in terms of the tangent matrix using ( [ ba4a ] ) . collecting all quadratic terms we get : @xmath104 \nonumber \\ = & \frac{1}{{m_+}^2 m_{qp } } \ , \frac{\delta { x_f}^2}{2b^2 } \left [ i m_{qq } -2i(m_{qq}+i m_{qp } ) + i m_{pp } { m_+}^2 - m_{qp } \right ] \nonumber \\ = & \frac{1}{{m_+}^2 m_{qp } } \ , \frac{\delta { x_f}^2}{2b^2 } \left [ -i m_{qq } + m_{qp } + i m_{pp } { m_+}^2 \right ] \nonumber \\ = & -\frac{i}{{m_+ } m_{qp } } \ , \frac{\delta { x_f}^2}{2b^2 } \left [ 1-{m_+ } m_{pp } \right ] = -\frac{\delta { x_f}^2}{2 b^2 } \ , \frac{m_{pp}-im_{pq}}{m_{qq}+im_{qp}}\end{aligned}\ ] ] where in the last equality we have used @xmath105 . with these simplifications the semiclassical wavepacket formula becomes @xmath106 \;.\end{gathered}\ ] ] the subscript @xmath107 on @xmath70 indicates the variables used to compute the real trajectory . the spreading of the propagated wavepacket is now explicit in the coefficient of the gaussian term . this is exactly heller s thawed gaussian approximation or tga @xcite . as discussed at the end of section 7 , a similar approximation @xcite can be obtained with coherent state path integrals . we consider now as 0-order the real trajectory which starts at @xmath12 and ends at @xmath5 after time @xmath6 . we call @xmath109 its initial momentum , which is a function of @xmath5 , @xmath12 , and @xmath6 , and we write @xmath110 notice that the complete expansion of @xmath62 to first order should be @xmath111 but , as @xmath5 is fixed , @xmath112 . eq.([ba9x ] ) gives us the relation between @xmath98 and @xmath113 @xmath114 thanks to eqs.([deltap1 ] ) and ( [ deltap2 ] ) we obtain @xmath115 which gives , in terms of the tangent matrix , @xmath116 we expand the exponent of eq.([ba14 ] ) around this real trajectory . once again we take the action s , its derivatives @xmath117 , @xmath118 , and its second derivatives @xmath119 , @xmath120 , @xmath121 , including the @xmath28 matrix , for the 0-order trajectory , unless we state otherwise explicitly . we obtain @xmath122 finally , the semiclassical propagator as a function of @xmath123 , and @xmath109 , becomes @xmath124 \ ; . \label{wv2}\end{gathered}\ ] ] the gaussian in the exponent is now in the difference between @xmath109 , the initial momentum of the real trajectory , and @xmath13 , the initial momentum of the center of the wave packet . notice that there might be more than one trajectory satisfying the boundary conditions @xmath125 , @xmath126 notice also that the wave function is not restricted to a gaussian anymore . and that this is _ not _ an initial value formula . in the same way , if we choose as 0-order the trajectory specified by @xmath13 and @xmath5 instead of @xmath12 and @xmath5 , we call @xmath128 the initial position for this trajectory , which is a function of @xmath5 , @xmath13 , and @xmath6 , and we write @xmath129 thanks to eqs.([ba9x ] ) and ( [ defs3 ] ) the new expression for @xmath98 is @xmath130 in the end we obtain a propagating semiclassical wavepacket different from the other two , namely @xmath131 \ ; . \label{wv3}\end{gathered}\ ] ] the gaussian in the exponent is in the difference between @xmath128 , the initial position of the real trajectory , and @xmath12 , the initial position of the center of the wavepacket . this formula shares many of the features of the previous one , including not being a gaussian and not being an initial value formula . the three real trajectory approximations we derived here are formally similar . among the three real variables @xmath12 , @xmath13 , and @xmath5 , two are chosen to fix the trajectory . then the corresponding formula carries a gaussian damping factor in the third variable . in the next three sections we discuss some examples that will help us compare these approximations with @xmath132 which involves complex trajectories . as usual all these formulae are exact for the free particle and the harmonic oscillator . the exact result for the propagation of a free particle wavepacket is of course well - known , but it can serve as a test of the formulae of secs . 2 and 3 . and it is instructive to be able to see explicitly the different trajectories involved in each approximation . the wave function for the packet at time 0 is given by ( [ ba6 ] ) . the hamiltonian is simply @xmath133 where @xmath16 is the mass . the action is @xmath134 all trajectories have constant momentum or velocity . in addition to the other parameters we shall use @xmath135 , which is the frequency of the oscillator upon which the coherent states are built , and @xmath136 , the central velocity of the wavepacket . the elements of the tangent matrix are @xmath137 . the exact expression for the propagated wavepacket follows from elementary quantum mechanics and can be written @xmath138 . \label{freewp}\end{gathered}\ ] ] it describes a gaussian packet of constant momentum @xmath13 , whose real width @xmath139 increases with time , but whose total width is actually complex and given by @xmath140 for the approximation of sec.2 , the stationary point @xmath51 of the integration , which is the origin of the complex trajectory , and its associated momentum @xmath62 , which is the constant momentum of the trajectory , turn out to be @xmath141 one can check the two boundary conditions ( [ bcf ] ) @xmath142 one can also verify that , when @xmath5 has the value @xmath143 , one finds the simple answers @xmath144 and @xmath145 : the trajectory is just the central trajectory of the packet . let us look also at the three approximations with real trajectories . for the @xmath85 case , formula ( [ ba18 ] ) contains both @xmath146 and @xmath147 the former is clearly @xmath148 , while @xmath149 , the constant momentum . for @xmath86 , formula ( [ wv2 ] ) , we need @xmath109 which is @xmath150 . we do nt need @xmath151 , but it is obviously the same as @xmath109 . finally , for @xmath87 , formula ( [ wv3 ] ) , we need @xmath152 since the constant momentum is @xmath153 once again , for the free particle , all four semiclassical approximations give the exact answer . to make things less trivial , we shall now have the wavepacket bounce elastically against a hard wall placed at the origin . the packet arrives from the right with a negative momentum and we assume that it starts far enough to have no appreciable value at the wall at time 0 . the only relevant spatial region is @xmath154 the other side of the wall does not exist for this problem . of course the wall could be an approximation to a very steep , but not totally hard , potential , in which case there would be some barrier penetration to the other side . and it would be interesting to see what happens in the limiting process , but we shall not do that here . the exact solution is quite simple , once one knows the exact free particle formula ( [ freewp ] ) : you take the @xmath155 part of the latter , you reflect it with respect to the wall , and you give it an additional minus sign . thus the complete solution has two parts , both restricted to @xmath156 the original free particle part , and the reflected part which uses the @xmath155 piece of the free particle : @xmath157 this construction ensures that the schroedinger equation is satisfied everywhere and that the total wave function vanishes at the wall . next we look at the semiclassical approximation with complex trajectories , sec . 2 . given @xmath2 and @xmath5 , both @xmath158 , there are two ways to go from one to the other in time @xmath6 , the direct way and the way which bounces off the wall . the first trajectory is identical in all respects to a free particle s : it has the same action ( [ fpa ] ) and the process of finding the stationary point @xmath51 of the integration over @xmath2 is the same . hence , for this direct trajectory , formula ( [ ba14 ] ) yields the free particle result , restricted to @xmath159 the reflected trajectory , on the other hand , has a different action . the distance between @xmath2 and @xmath5 , including reflection , is @xmath160 , the speed is @xmath161 , the momentum @xmath162 , and the energy @xmath163 . hence the action is @xmath164 it differs from ( [ fpa ] ) by a sign change for one of the two coordinates . the initial and final momenta follow from this action in the usual way : @xmath165 compared to the direct trajectory , the only change that needs to be made when we look for the stationary point of the @xmath2 integration and we apply formula ( [ ba14 ] ) , is that the constant quantity @xmath5 must be replaced everywhere by @xmath166 . thus we get again the free particle result , but for the value @xmath166 of the position . this approximation , therefore , yields the exact answer ( [ wall ] ) , except for the minus sign in front of the second term . this minus sign due to reflection is expected on very general grounds . it is a special case of the morse phase mentioned in sec . 2 , and it is understood easily if one looks at a soft barrier , such as a finite square step , and one goes to the infinite limit . now let us try to apply to the hard wall the approximation with the @xmath85 real trajectory of sec . we shall see that it does not work for this kind of problem . in this formulation there is a single trajectory , which begins at @xmath85 and evolves with @xmath6 . its endpoint is given by @xmath167 until it hits the wall , which happens for time @xmath168 . after this time @xmath146 becomes @xmath169 . formula ( [ ba18 ] ) yields for a given @xmath6 a single gaussian wavepacket whose center follows this trajectory . the packet is incoming for @xmath170 and outgoing for @xmath171 . incoming and outgoing are never present at the same time and therefore there are no interference effects . in a calculation for a soft wall , one would see the full amplitude of the gaussian packet on the back side of the wall , with no damping due to barrier penetration , and again no interference effects . on the other hand , the other two approximations with real trajectories , @xmath86 and @xmath87 , do work fine for the hard wall and give the exact answer . each of them contains a direct and a reflected trajectory at all times , hence a sum of two wavepackets with interference . the hard wall is probably the simplest example after the free particle . however , except for @xmath172 , the calculation involves two trajectories . our aim here is to consider a potential where a single classical trajectory contributes to the semiclassical formulae , to avoid dealing with interference and focus only on the differences arising from the complex or real character of a trajectory . we choose an inverted gaussian potential with hamiltonian @xmath173 the system has a single parameter , which is the width of the wavepacket . we set @xmath174 and we place the packet initially at @xmath175 , @xmath176 , hence central energy 1/8 . there is a single bound level with energy -0.48 , which excludes two simple limits : that in which the potential is just a perturbation , and that in which the problem is highly semiclassical . here both the real and the complex trajectories need to be calculated numerically , as well as the exact packet propagation of course . the numerical methods for the trajectories will be discussed in the next section . 1 shows the wave packet at @xmath177 for the different approximations , compared with the exact result , for two values of the width , @xmath178 and @xmath179 the packet spreads very fast and rapidly becomes highly non - gaussian . it is remarkable how well this non - gaussian character is reproduced by some of the approximations , though not all . we shall have more to say in section 7 about comparing the various approximations . the top two graphs are the probability density versus @xmath5 . the bottom two are the phase of the wave function . both the modulus and the phase are very well reproduced by @xmath180 the approximation via complex trajectories , and @xmath181 one of the real trajectories approximations , @xmath182 being just as good as @xmath132 for this . on the other hand @xmath172 , which is a pure gaussian , does not do well at all . the small zig - zags in the approximate calculations of the phase are due to numerical imprecisions . here , as well as in the next section , we do not show @xmath183 because it displays basically the same features as its counterpart @xmath184 calculations with other values of @xmath41 show these results to be robust , in the sense that @xmath182 is always very similar to @xmath132 and that both agree well with the exact propagation . in this section we apply our semiclassical formulae to the case of a totally binding potential . we choose the quartic oscillator because it is probably the simplest system after the harmonic oscillator ( for which all semiclassical formulae of sections 2 and 3 are exact ) . the hamiltonian is @xmath185 and the parameters are set to @xmath186 , @xmath187 , @xmath188 for these values the ground state energy is @xmath189 and the first two excited states have @xmath190 and @xmath191 . for the wavepacket we choose @xmath192 , @xmath193 , and @xmath194 this gives @xmath195 for the energy of the central trajectory , @xmath196 for its period , and @xmath197 for its turning points . fig.2 shows a sequence of four snapshots in the exact time evolution of the packet . the energy is low , the wavelength is large , and the interference effects are important . as in the previous example , both the real and the complex trajectories have to be computed numerically . the real trajectories from @xmath12 to @xmath5 can be calculated with a simple ` shooting method ' . since we need the propagated wave packet at all values of @xmath5 , we simply integrate the equations of motion starting from @xmath125 ( fixed ) and several values of the initial momentum @xmath198 . for each @xmath109 we obtain a final coordinate @xmath199 at which we evaluate the wave function . the calculation of the complex trajectories is more involved and we shall describe it in some detail . we follow the procedure introduced by klauder @xcite ( see also @xcite and @xcite for a different approach ) , which consists in propagating trajectories starting from @xmath200 where @xmath201 is a complex number . the first of conditions ( [ bcf ] ) is automatically satisfied for all @xmath202 . the search method consists in propagating trajectories for all possible @xmath202 s and picking those satisfying the second condition ( [ bcf ] ) , @xmath126 as discussed in @xcite , fixing @xmath12 , @xmath13 and @xmath6 and integrating hamilton s equations with initial conditions ( [ s42 ] ) gives , for each @xmath202 , a final coordinate @xmath203 , usually complex . this can be viewed as a map @xmath204 . the values of @xmath202 we need are given by the inverse map of the real @xmath205 axis . if the map is analytic at @xmath202 , then it is conformal . this implies , among other things , that the map is one - to - one in the neighborhood of @xmath202 . if , on the other hand , @xmath202 is a critical point of the map , a richer structure develops . if , for instance , @xmath206 , but the second derivative is non - zero , it can be shown that the map becomes two - to - one in the vicinity of @xmath202 , meaning that two different trajectories ( corresponding to two distinct initial conditions ) satisfy the same boundary conditions . this is the basic mechanism that generates multiple trajectories in systems with one degree of freedom . for short times the critical points of the map generically lie very far from the origin @xmath207 , corresponding to complex trajectories whose actions have large imaginary parts . as time increases , more and more of these points approach the origin , giving rise to significant contributions to the propagator @xcite . the singularities in the @xmath208 map produced by the critical points are called phase space caustics ( psc ) . at these points the square - root in the pre - factor of eq.([ba14 ] ) goes to zero and the semiclassical approximation fails ( it actually fails in a finite neighborhood of the psc s ) . for fixed @xmath209 , the complex trajectories form continuous families parameterized by @xmath210 among the many families that might contribute to the semiclassical evolution of the wave - packet , one is special . this is the family that contains the real trajectory that starts at @xmath85 , and we call it the _ main family_. if @xmath211 is the end point of this real trajectory , then for @xmath212 the trajectory in the main family is real . in terms of the map , the point @xmath207 is mapped into @xmath213 . in some special situations , the main family alone may provide sufficient information for the semiclassical calculation @xcite , but that is not always the case @xcite . fig.3(a ) shows the map @xmath214 for @xmath215 . the lines correspond to constant values of @xmath216 and @xmath217 . the conformal property of the map guarantees that these lines cross at right angle ( provided the same scale is used for @xmath218 and @xmath219 ) . the circle indicates the location of a critical point , corresponding to a singularity in the regular pattern of the lines . the thick lines correspond to families of trajectories for which @xmath220 the main family can be identified by its containing the point @xmath221 a more careful look at this figure reveals the existence of several other singularities in the otherwise regular pattern of crossing lines . each of them is a critical point of the map . the trajectories in their neighborhood , however , do not contribute significantly to the propagated wave packet , and we shall not take them into account . let @xmath222 be the exponent in eq.([ba14 ] ) . for real trajectories @xmath223 is @xmath224 . fig.3(b ) shows a greyscale topographic map of @xmath225 for the trajectories calculated in fig.3(a ) and in the same @xmath226 plane . the continuous thick line is the main family seen on fig.3(a ) . the line for the other family , which we call _ secondary _ , is part thick and part thin . the reason for this distinction is the following . the imaginary part of @xmath227 for trajectories in the main family can be seen to be always positive . it starts with @xmath228 for @xmath229 , decreases to zero at @xmath212 and grows to infinity again as @xmath230 . for the trajectories in the secondary family , however , @xmath225 starts at @xmath231 for @xmath232 , and decreases steadily to @xmath233 at @xmath234 . this means that this family can not be included in the semiclassical calculation for all values of @xmath5 . there must be a cutting point after which the family can not be considered , otherwise it would cause the wave function to diverge . this phenomenon has become known as the _ problem of non - contributing trajectories _ , and has been studied by adachi @xcite , berry @xcite , klauder @xcite and others @xcite . the part of the secondary family shown with the thin line has not been included in our calculations . we note that , in @xcite , huber and heller do a similar calculation for a wave packet moving in the morse potential . to understand qualitatively the role of secondary families in the semiclassical limit , we consider @xmath26 to be really very small . in this case the @xmath86 and @xmath87 real trajectory approximations of section 3 become exact , and only the main family ( actually only a small neighborhood of the real trajectory ) contributes significantly . demanding the approximation to be uniformly valid as @xmath235 , and using the subscript @xmath28 for the main family and @xmath236 for a secondary one , we must apply the following rules : ( a ) trajectories with @xmath237 should be removed @xcite . this avoids the divergence of @xmath238 in the limit @xmath239 ; ( b ) trajectories with @xmath240 but @xmath241 should also be removed to guarantee that , as @xmath239 , the main family always gives the dominant contribution ; and ( c ) the discontinuity introduced by the sudden removal of a secondary contribution should be minimized . this criterion ultimately determines the choice of the cutoff point , which is located on the so - called stokes line @xcite . fig.4 shows the comparison between the exact calculation ( thick line ) and the different approximations of sections 2 and 3 for @xmath242 and @xmath243 the most interesting feature here is that the real trajectory formula @xmath182 becomes discontinuous . this can be understood with the help of panels ( b ) and ( d ) which show the initial momentum @xmath109 of the contributing trajectories as a function of @xmath5 ( for fixed @xmath12 and @xmath6 ) . we see that several branches appear , but it turns out that only the one shown with a thick line contributes significantly . since this main branch covers only a finite range of @xmath5 , the wave function drops to zero suddenly . in fact , the wave function diverges at the ends of the branches . in our calculations we have cut the branch a little before its end points . fig.5(a ) shows the real part of @xmath244 for @xmath215 calculated with the separate contribution of each of the families shown in figs.3 . the abrupt cutoff of the secondary family is clear in this figure . fig.5(b ) shows again the separate contributions of the same families to @xmath245 ( dashed and solid lines ) and their combined contribution ( thick solid line ) . notice the interference between the two families producing the oscillation in the probability density . finally fig.6 shows @xmath245 , the calculation with complex trajectories . each snapshot shows the exact result ( thin solid line ) , the thawed gaussian approximation @xmath172 ( dashed line ) and the complex trajectories approximation ( thick line ) . we do not show @xmath182 in these plots because the approximation is not good . the improvement obtained with the complex trajectories is clear . for @xmath215 and @xmath246 , however , we can see a spurious peak close to the turning points . this might seem to result from cutting off the secondary family at the wrong point , as suggested in @xcite . a careful analysis shows that it is not the case : cutting off the secondary family at a smaller value of @xmath5 would produce a sudden dip in the probability density followed by another sudden increase , as can be seen from fig.5(b ) . the spurious peaks are due to the phase space caustic that shows up close to the turning points . for the present situation , which is not quite in the semiclassical domain , the caustics have a strong effect on the semiclassical propagations and further corrections are necessary to improve the approximation . a uniform semiclassical approximation involving airy functions can be derived by considering cubic terms in the saddle point approximation performed in the integral ( [ ba7 ] ) . the derivation of this improved formula will be published elsewhere . we begin by summarizing the results . we have derived and written down in this paper ( sections 2 and 3 ) four semiclassical expressions for the propagation of an initially gaussian wavepacket , @xmath247 all four give exact results for the free particle and the harmonic oscillator . all except @xmath172 also give the exact result for the scattering by a hard wall . but @xmath172 is very bad for the hard wall : it produces no interference . we tested the formulae further on two examples of smooth potentials . for the attractive gaussian potential , the exact packet after some time is highly non - gaussian . both @xmath248 and @xmath182 involve a single trajectory for each @xmath5 , and both give very good results for the modulus as well as the phase of the wave packet . the same is true of @xmath249 but @xmath250 which is a pure gaussian , gives a very bad fit . the other smooth potential is a quartic oscillator . now there may be contributions from multiple families of trajectories and the results are quite different . we found in this case that the complex trajectories formula @xmath248 is superior to both the real trajectories approximation @xmath182 and the single trajectory approximation @xmath172 . one reason is that the real trajectories contributing to @xmath182 have finite branches , which causes a drastic discontinuity in the approximate wave function . the complex trajectories , on the other hand , form continuous branches that prolong into the complex domain . the projection of these complex families onto the real plane show that they follow closely the main branch of the real trajectories , but they continue to exist after the latter end . thus the paper s main finding is that the complex trajectory approximation , @xmath248 of eq . ( [ ba14 ] ) , gives very good results , at least for all the cases that we looked at . and one may well wonder how it could be that a relatively simple approximation , based on a single trajectory , works so well . the short answer is that , actually , the approximation is not that simple and it is not based on a single trajectory . but we shall go into more details . many other approaches have been proposed for the propagation of an initial wave function , some of them especially concerned with the propagation in systems with many degrees of freedom . among them , initial value representations or ivr s @xcite have become especially popular . they are integral expressions over the phase space of initial conditions , in a spirit which tends to rejoin that of path integrals . the multiple integration is usually handled by monte carlo techniques @xcite . the initial impetus for these approaches @xcite was the root - search problem , i.e. the fact that the van vleck expression ( [ ba5 ] ) requires one to find trajectories determined by mixed initial - final boundary conditions ( starting at @xmath2 and ending at @xmath5 ) . this may lead to a difficult search in complicated situations , and in addition the root - search problem usually has several solutions , a fact which complicates further the time - development of the wave function . in the ivr s , on the other hand , the trajectories are determined solely by initial boundary conditions , as the name implies , and for each set of initial conditions there is only one trajectory , the unique solution of hamilton s equations for these conditions . the work of getting the correct propagated wave function is done entirely by the integration , as is the case for path integral expressions . the present work is based on the van vleck expression and it is affected by root - search , except for the thawed gaussian formula @xmath172 of section 3.1 . root - search can truly be a burden , especially for multi - dimensional wave functions . more important , however , is the fact that characterizing the present approach as a one - trajectory or a few - trajectories approximation is highly misleading . we have used such characterization several times above , for instance in connection with the hard wall , or in the first few lines of section 5 . but the truth is that there is a different classical trajectory for every final @xmath5 , and therefore the number of trajectories is really infinite . the work of getting the correct propagated wave function is done by them . the one case where we truly have a one - trajectory expression is @xmath251 there , the same trajectory is used for all values of @xmath5 . this case is truly an ivr , though a very simplified one , not involving any integration over the initial parameters . the only initial parameters used are those of the center of the packet . and it is also the case for which , in our explorations , the agreement with the exact result is worst , showing that a single classical trajectory is just not enough . somehow , a variety of classical trajectories must enter into the formalism . in other words , we see the path integral idea reappear , feynman s original view of the connection between quantum and classical mechanics . among the various ivr s , there is one that has been especially successful and has had many applications , the herman - kluk approximation @xcite . it would be particularly interesting to have a direct application of the hk formula to the hard wall , since this is the simplest non - obvious case for which our complex trajectories formula is exact . we are not aware of the existence of such an application of hk , and it seems to us that the calculation will not be totally straightforward . perhaps it is not superfluous to repeat once again that the main new result of this paper is that semiclassical approximations using complex trajectories , especially non - ivr approximations , are capable of being surprisingly good . all our semiclassical expressions generalize easily to many dimensions . the detailed presentation of these generalizations will be published separately , together with numerical applications . we note that two - dimensional semiclassical calculations with complex trajectories have already been considered in the context of coherent states . in @xcite the quantity evaluated was the return probability amplitude , also known as the auto - correlation function . in @xcite the authors , who seem to have been unaware of references @xcite , calculate the propagated wave - function in the momentum or coordinate representation . but they do it for quantum maps , which is rather different from what we have done here since , in their case , there are no continuous classical trajectories , no hamilton equations , and no schroedinger equation . in the present paper we provided formulae for the calculation of the propagated wave - function for continuous systems in the coordinate representation . the basic equation for such calculations is the complex trajectory formula ( [ ba14 ] ) first derived by heller and coworkers in @xcite . we showed that one can approximate the complex trajectories by real ones , and that this can be done in various ways ( we mention three of them ) , with different degrees of difficulty and of accuracy . the various formulae so generated will undoubtedly have different domains of applicability . and the complex formula , if it is practical , will always remain the better one . as our final topic , we shall present an alternative derivation of a formula very similar to ( [ ba14 ] ) , which possesses the same three approximations in terms of real trajectories . the quantity we want to calculate is @xmath252 . to obtain the approximation ( [ ba3a ] ) , we introduced between @xmath8 and @xmath15 the resolution of unity in terms of eigenstates of @xmath253 , and then we replaced the exact @xmath253-propagator by its van vleck approximation . instead of this , we shall introduce between @xmath254 and @xmath8 the resolution of unity in terms of coherent states , thus @xmath255 and then we shall replace the exact coherent state propagator by its semiclassical approximation . the latter was worked out in detail in @xcite . performing the integrals over @xmath256 and @xmath257 by the stationary exponent approximation , we arrive at an expression which is very similar , but not identical , to ( [ ba14 ] ) . there are two important differences . one is that , in this case , the dynamics are governed by the smoothed hamiltonian @xmath258 instead of the classical hamiltonian . the other is that the exponent acquires an extra term given by @xcite @xmath259 this term plays a very important role and can not be dropped from the approximation . both @xmath260 and the @xmath261 dynamics get carried over into all the real trajectories approximations . our main reason for mentioning this alternative is to stress that semiclassical approximations are never unique and that there are always many ways to base an approximate quantal expression on a classical solution . discussions of this appear in @xcite , who see it as a consequence of the over - completeness of the coherent - state basis , and also in @xcite . it is only in the limit of @xmath26 going to zero that they all become the same . from our alternative to ( [ ba14 ] ) involving the smoothed hamiltonian and the correction term @xmath260 , one can derive again three expressions in terms of real trajectories , different from those in section 3 . numerically , the two points of view yield roughly equally good ( or sometimes equally bad ) results . and once again , for quadratic hamiltonians , all the approximations we have mentioned are exact . while in section 3 the @xmath172 approximation was the same as heller s tga , for the alternative formulation it is the same as the approximation in section 4 of @xcite . the latter reference contains a discussion of the differences between the two results . both are initial value formulae using the real central trajectory of the packet , and both are gaussian in shape for all potentials and all times . both are expected to fail if the potential is not very smooth on the scale of the size of the packet . + acknowledgements	we consider a semiclassical approximation , first derived by heller and coworkers , for the time evolution of an originally gaussian wave packet in terms of complex trajectories . we also derive additional approximations replacing the complex trajectories by real ones . these yield three different semiclassical formulae involving different real trajectories . one of these formulae is heller s thawed gaussian approximation . the other approximations are non - gaussian and may involve several trajectories determined by mixed initial - final conditions . these different formulae are tested for the cases of scattering by a hard wall , scattering by an attractive gaussian potential , and bound motion in a quartic oscillator . the formula with complex trajectories gives good results in all cases . the non - gaussian approximations with real trajectories work well in some cases , whereas the thawed gaussian works only in very simple situations .
a vast number of astronomical observations suggests that magnetic fields play a crucial role in the dynamics of many phenonema of relativistic astrophyics , either on stellar scales , such as for pulsars , magnetars , compact x - ray binaries , short and long / gamma - ray bursts ( grbs ) and possibly for the collapse of massive stellar cores , but also on much larger scales , as it is the case for radio galaxies , quasars and active galactic nuclei ( agns ) . a shared aspect in all these phenomena is that the plasma is essentially electrically neutral and the frequency of collisions is much larger than the inverse of the typical timescale of the system . the mhd approximation is then an excellent description of the global properties of these plasmas and has been employed with success over the several decades to describe the dynamics of such systems well in their nonlinear regimes . another important common aspect in these systems is that their flows are characterized by large magnetic reynolds numbers @xmath0 , where @xmath1 and @xmath2 are the typical sizes and velocities , respectively , while @xmath3 is the magnetic diffusivity and @xmath4 is the electrical conductivity . for a typical relativistic compact object , @xmath5 and , under these conditions , the magnetic field is essentially advected with the flow , being continuosly distorted and possibly amplified , but also essentially not decaying . we note that these conditions are very different from those traditionally produced in the earth s laboratories , where @xmath6 , and the resistive diffusion represents an important feature of the magnetic - field evolution . a particularly simple and yet useful limit of the mhd approximation is that of the _ `` ideal - mhd '' _ limit . this is mathematically defined as the limit in which the electrical resistivity @xmath7 vanishes or , equivalently , by an infinite electrical conductivity . it is within this framework that many multi - dimensional numerical codes have been developed over the last decade to study a number of phenomena in relativistic astrophysics and in fully nonlinear regimes @xcite . the ideal - mhd approximation is not only a convenient way of writing and solving the equations of relativistic mhd , but it is also an excellent approximation for any process that takes place over a dynamical timescale . in the case of an old and `` cold '' neutron star , for example , the electrical and thermal transport properties of the matter are mainly determined by the transport properties of the electrons , which are the most important carriers of charge and heat . at temperatures above the crystallization temperature of the ions , the electrical ( and thermal ) conductivities are governed by electron scattering off ions and an approximate expression for the electrical conductivity is given by @xcite @xmath8 , where @xmath9 and @xmath10 are the stellar temperature and mass density g @xmath11 and temperatures in the range @xmath12 k , but provides a reasonable estimate also at larger temperatures of @xmath13 k [ _ cf . @xcite ] . ] . even for a magnetic field that varies on a length - scale as small as @xmath14 , ( where @xmath15 is the stellar radius ) the magnetic diffusion timescale is @xmath16 . clearly , at these temperatures and densities , ohmic diffusion will be neglible for any process taking place on a dynamical timescale for the star , _ i.e. _ , @xmath17 , and thus the conductivity can be considered as essentially infinite . however , catastrophic events , such as the merger of two neutron stars , or of a neutron star with a black hole , can produce plasmas with regions at much larger temperatures ( _ e.g. _ , @xmath18 ) and much lower densities ( _ e.g. _ , @xmath19 . in such regimes , all the transport properties of the matter will be considerably modified and non - ideal effects , absent in perfect - fluid hydrodynamics ( such as bulk viscosity ) and ideal mhd ( such as ohmic diffusion on a much shorter timescale @xmath20 ) will need to be taken into account . similar conditions are likely not limited to binary mergers but , for instance , be present also behind processes leading to long grbs , thus extending the range of phenomena for which resistive effects could be important . note also that these non - ideal effects in hydrodynamics ( mhd ) are proportional not only to the viscosity ( resistivity ) of the plasma , but also to the second derivatives of the velocity ( magnetic ) fields . hence , even in the presence of a small viscosity ( resistivity ) , their contribution to the overall conservation of energy and momentum can be considerable if the velocity ( magnetic ) fields undergo very rapid spatial variations in the flow . a classical example of the importance of resistive mhd effects in plasmas with high but finite conductivities is offered by _ current sheets_. these phenomena are often observed in the solar activity and are responsible for the reconnection of magnetic field lines and changes in the magnetic field topology . while these phenomena are behind the emission of large amounts of energy , they are strictly forbidden within the ideal - mhd limit due to magnetic flux conservation and so can not be studied employing this limit . besides having considerably smaller conductivities , low - density higly magnetized plasmas are present rather generically around magnetized objects , constituting what is referred to as the `` magnetosphere '' . in such regions magnetic stresses are much larger than magnetic pressure gradients and can not be properly balanced ; as a result , the magnetic fields have to adjust themselves so that the magnetic stresses vanish identically . this scenario is known as the _ force free _ regime ( because the lorentz force vanishes in this case ) and while the equations governing it can be seen as the low - inertia limit of the ideal - mhd equations @xcite , the force - free limit is really distinct from the ideal - mhd one . this represents a considerable complication since it implies that it is usually not possible to decribe , within the same set of equations , both the interior of compact objects and their magnetospheres . theoretical work to derive a fully relativistic theory of non - ideal hydrodynamics and non - ideal mhd has been carried out by several authors in the past @xcite and is particularly simple in the case of the resistive mhd description . the purpose of this work is indeed that of proposing the solution of the relativistic resistive mhd equations as an important step towards a more realistic modelling of astrophysical plasmas . there are a number of advantages behind such a choice . first , it allows one to use a single mathematical framework to describe both regions where the conductivity is large ( as in the interior of compact objects ) and small ( as in magnetospheres ) , and even the vacuum regions outside the compact objects where the mhd equations trivially reduce to the maxwell equations . second , it makes it possible to account self - consistently for those resistive effects , such as current sheets , which are energetically important and could provide a substantial modification of the whole dynamics . last but not least , the numerical solution of the resistive mhd equations provides the only way to control and distinguish the physical resistivity from the numerical one . the latter , which is inevitably present and proportional to truncation error , is also completely dependent on the specific details of the numerical algorithm employed and on the resolution used for the solution . as noted already by several authors , the numerical solution of the ideal - mhd equations is considerably less challenging than that of the resistive mhd equations . in this latter case , in fact , the equations become mixed hyperbolic - parabolic in newtonian physics or hyperbolic with stiff relaxation terms in special relativity . the presence of stiff terms is the natural consequence of the fact that the diffusive effects take place on timescales that are intrinsically larger than the dynamical one . stated differently , in such equations the relaxation terms can dominate over the purely hyperbolic ones , posing severe constraints on the timestep for the evolution . while considerable work has already been made to introduce numerical techniques to achieve efficient implementations in either regime @xcite , the use of these techniques in fully three - dimensional simulations is still difficult and expensive . in order to benefit from the many advantages discussed above in the use of the resistive mhd equations , we here present a novel approach for the solution of the relativistic resistive mhd equations exploiting the properties of implicit - explicit ( imex ) runge kutta methods . this approach represents a simple but effective solution to the problem of the vastly different timescales without sacrificing either simplicity in the implementation or the numerical efficiency . by examining a number of tests we illustrate the accuracy of our approach under a variety of conditions and demonstrate its robustness . in addition , we also compare it with the alternative method proposed by @xcite for the solution of the same set of relativistic resistive mhd equations . this latter approach employs strang - splitting techniques and the analytical integration of a reduced form of ampere s law . while it works well in a number of cases , it has revealed to be unstable when applied to discontinuous flows with large conductivities ; such difficulties were not encountered when solving the same problem within the imex implementation . because our approach effectively treats within a unified framework both those regions of the flow which are fluid - pressure dominated and those which are instead magnetic - pressure dominated , it could find a number of applications and serve as a first step towards a more realistic modeling of relativistic astrophysical plasmas . our work is organized as follows . in sect . [ section2 ] we present the system of equations describing a resistive magnetized fluid , while in section [ section3 ] we discuss the problems related to the numerical evolution of this system of equations and the numerical approaches developed to solve them . in particular , we introduce the basic features of the imex runge - kutta schemes and recall their stability properties . in sect [ section4 ] we instead explain in detail the implementation of the imex scheme to the resistive mhd equations . finally , in sect . [ section5 ] we present the numerical tests carried out either in one or two dimensions and that span several prescriptions for the conductivity . section [ section5 ] is also dedicated to the comparison with the strang - splitting technique . the conclusions and the perspectives for future improvements are presented in sect . [ section6 ] , while appendix [ appendixb ] reviews our space discretization of the equations . hereafter we will adopt gaussian units such that @xmath21 and employ the summation convention on repeated indices . roman indices @xmath22 are used to denote spacetime components ( _ i.e. _ , from @xmath23 to @xmath24 ) , while @xmath25 are used to denote spatial ones ; lastly , bold italics letters represent vectors , while bold letters represent tensors . an effective description of a fluid in the presence of electromagnetic fields can be made by considering three different sets of equations governing respectively the electromagnetic fields , the fluid variables and the coupling between the two . in particular , the electromagnetic part can be described via the maxwell equations , while the conservation of energy and momentum can be used to express the evolution of the fluid variables . finally , ohm s law , whose exact form depends on the microscopic properties of the fluid , expresses the coupling between the electromagnetic fields and the fluid variables . in what follows we review these three sets of equations separately , discuss how they then lead to the resistive mhd description , and how the latter reduces to the well - known limits of ideal - mhd and of the maxwell equations in vacuum . our presentation will be focussed on the special - relativistic regime , but the extension to general relativity is rather straightforward and will be presented elsewhere . the special relativistic maxwell equations can be written as @xcite @xmath26 where @xmath27 and @xmath28 are the maxwell and the faraday tensor respectively and @xmath29 is the electric current 4-vector . a highly - ionized plasma has essentially zero electric and magnetic susceptibilities and the faraday tensor is then simply the dual of the maxwell tensor . this tensor provides information about the electric and magnetic fields measured by an observer moving along any timelike vector @xmath30 , namely @xmath31 we are considering @xmath30 to be the time - like traslational killing vector field in a flat ( minkowski ) spacetime , so @xmath32 and the levi - civita symbol @xmath33 is non - zero only for spatial indices . note that the electromagnetic fields have no components parallel to @xmath30 ( _ i.e. _ , @xmath34 ) . by using the decomposition of the maxwell tensor ( [ maxwell_tensor ] ) , the equations ( [ maxwell_covariant1])([maxwell_covariant2 ] ) can be split into directions which are parallel and orthogonal to @xmath30 to yield the familiar maxwell equations @xmath35 where we have decomposed also the current vector @xmath36 , with @xmath37 being the charge density , @xmath38 the convective current and @xmath39 the conduction current satisfying @xmath40 . the current conservation equation @xmath41 follows from the antisymmetry of the maxwell tensor and provides the evolution of the charge density @xmath37 @xmath42 which can be obtained also directly by taking the divergence of ( [ maxwell_clasic3 ] ) when the constraints ( [ maxwell_clasic1])([maxwell_clasic2 ] ) are satisfied . the evolution of the matter follows from the conservation of the stress - energy tensor @xmath43 and the conservation of baryon number @xmath44 where @xmath10 is the rest - mass density ( as measured in the rest frame of the fluid ) and @xmath45 is the fluid 4-velocity . the stress - energy tensor @xmath46 describing a perfect fluid minimally coupled to an electromagnetic field is given by the superposition @xmath47 where @xmath48 here @xmath49 is the enthalpy , with @xmath50 the pressure and @xmath51 the specific internal energy . the conservation law ( [ conservation_tmunu ] ) can be split into directions parallel and orthogonal to @xmath30 to yield the familiar energy and momentum conservation laws @xmath52 where we have introduced the conserved quantities @xmath53 , which are essentially the energy density @xmath54and the energy flux density @xmath55 , and whose expressions are given by @xmath56 here @xmath57 is the velocity measured by the inertial observer and @xmath58 is the lorentz factor . the fluxes can then be written as @xmath59 { \bf g } \ , . \label{def_fs } \end{aligned}\ ] ] finally , the conservation of the baryon number ( [ conservation_baryons ] ) reduces to the continuity equation written as @xmath60 where we have introduced another conserved quantity @xmath61 and its flux @xmath62 . as mentioned above , maxwell equations are coupled to the fluid ones by means of the current 4-vector @xmath63 , whose explicit form will depend in general on the electromagnetic fields and on the local fluid properties . a standard prescription is to consider the current to be proportional to the lorentz force acting on a charged particle and the electrical resistivity @xmath64 to be a scalar function . ohm s law , written in a lorentz invariant way , then reads @xmath65 with @xmath66 being the electrical conductivity of the medium . expressing ( [ ohm_relativistic_covariant ] ) in terms of the electric and magnetic fields one obtains the familiar form of ohm s law in a general inertial frame @xmath67 + q~{\boldsymbol v } \ , .\ ] ] note that the conservation of the electric charge ( [ current_conservation_clasic ] ) provides the evolution equation for the charge density @xmath37 ( _ i.e. _ , the projection of the 4-current @xmath68 along the direction @xmath69 ) , while ohm s law provides a prescription for the ( spatial ) conduction current @xmath70 ( _ i.e. _ , the components of @xmath71 orthogonal to @xmath69 ) . it is important to recall that in deriving expression ( [ ohm_relativistic ] ) for ohm s law we are implicitly assuming that the collision frequency of the constituent particles of our fluid is much larger that the typical oscillation frequency of the plasma . stated differently , the timescale for the electrons and ions to come into equilibrium is much shorter than any other timescale in the problem , so that no charge separation is possible and the fluid is globally neutral . this assumption is a key aspect of the mhd approximation . the well - known ideal - mhd limit of ohm s law can be obtained by requiring the current to be finite even in the limit of infinite conductivity ( @xmath72 ) . in this limit ohm s law ( [ ohm_relativistic ] ) then reduces to @xmath73 projecting this equation along @xmath57 one finds that the electric field does not have a component along that direction and then from the rest of the equation one recovers the well - known ideal - mhd condition @xmath74 stating that in this limit the electric field is orthogonal to both @xmath75 and @xmath57 . such a condition also expresses the fact that in ideal mhd the electric field is not an independent variable since it can be be computed via a simple algebraic relation from the velocity and magnetic vector fields . summarizing : the system of equations of the relativistic resistive mhd approximation is given by the constraint equations ( [ maxwell_clasic1])([maxwell_clasic2 ] ) , evolution equations ( [ maxwell_clasic3])([current_conservation_clasic ] ) , ( [ fluid_tau])([fluid_s ] ) and ( [ fluid_baryons ] ) , where the fluxes are given by eqs . ( [ def_fe])([def_fs ] ) and the 3-current is given by ohm s law ( [ ohm_relativistic ] ) . these equations , together with a equation of state ( eos ) for the fluid and a reasonable model for the conductivity , completely describe the system under consideration provided consistent initial and boundary data are defined . at this point it is useful to point out some properties of the relativistic resistive mhd equations discussed so far , to underline their purely hyperbolic character and to contrast them with those of other forms of the resistive mhd equations which contain a parabolic part instead . to do this within a simple example , we adopt the newtonian limit of ohm s law ( [ ohm_relativistic ] ) , @xmath76 \ , , \ ] ] where we have neglected terms of order @xmath77 , obtaining the following potentially stiff equation for the electric field @xmath78 \ , . \label{maxwell_stiff}\ ] ] assuming now a uniform conductivity and taking a time derivative of eq . ( [ maxwell_clasic4 ] ) , we obtain the following hyperbolic equation with relaxation terms ( henceforth referred simply as hyperbolic - relaxation equation ) for the magnetic field @xmath79 = [ \partial_t { \boldsymbol b } - \nabla \times ( { \boldsymbol v } \times { \boldsymbol b})]\,.\ ] ] if the displacement current can be neglected , _ i.e. _ , @xmath80 , equation ( [ relativistic_rmhd ] ) reduces to the familiar parabolic equation for the magnetic field @xmath81 where the last term is responsible for the diffusion of the magnetic field . it is important to stress the significant difference in the characteristic structure between equations ( [ relativistic_rmhd ] ) and ( [ newtonian_rmhd ] ) . both equations reduce to the same advection equation in the ideal - mhd limit of infinite conductivity ( @xmath82 ) indicating the flux - freezing condition . however , in the opposite limit of infinite resistivity ( @xmath83 ) eq . ( [ newtonian_rmhd ] ) tends to the ( physically incorrect ) elliptic laplace equation @xmath84 while eq . ( [ relativistic_rmhd ] ) reduces to the ( physically correct ) hyperbolic wave equation for the magnetic field . the set of maxwell equations described above can also be cast in an extended fashion which includes two additional fields , @xmath85 and @xmath86 , introduced to control dynamically the constraints of the system , _ i.e. _ , eqs ( [ maxwell_clasic1 ] ) and ( [ maxwell_clasic2 ] ) . this _ `` augmented '' _ system reads @xmath87 clearly , the standard maxwell equations ( [ maxwell_covariant1])([maxwell_covariant2 ] ) are recovered when @xmath88 and we are in this way extending the space of solutions of the original maxwell equations to include those with non - vanishing @xmath89 . the evolution of these extra scalar fields can be obtained by taking a partial derivative @xmath90 of the augmented maxwell equations ( [ maxwell_augmented1])([maxwell_augmented2 ] ) and using the antisymmetry of the maxwell and faraday tensors together with the conservation of charge to obtain @xmath91 it is evident that these represent wave equations with sources for the scalar fields @xmath89 , which propagate at the speed of light while being damped if @xmath92 . in particular , for any positive @xmath93 , they decay exponentially over a timescale @xmath94 to the trivial solution @xmath88 and the augmented system then reduces to the standard maxwell equations , including the constraints ( [ maxwell_clasic1 ] ) and ( [ maxwell_clasic2 ] ) . this approach , named hyperbolic divergence cleaning in the context of ideal mhd @xcite , was proposed as a simple way of solving the maxwell equations and enforcing the conservation of the divergence - free condition for the magnetic field . adopting this approach and following the formulation proposed by @xcite , the evolution equations of the augmented maxwell equations ( [ maxwell_augmented1])([maxwell_augmented2 ] ) can then be written as @xmath95 the system of equations ( [ maxwell_aug1])([maxwell_aug4 ] ) , together with the current conservation ( [ current_conservation_clasic ] ) , is the one we will use for the numerical evolution of the electromagnetic fields within the set of relativistic resistive mhd equations . while the ideal - mhd equations are well suited to an efficient numerical implementation , the general system of relativistic resistive mhd equations brings about a delicate issue when the conductivity in the plasma undergoes very large spatial variations . in the regions with high conductivity , in fact , the system will evolve on timescales which are very different from those in the low - conductivity region . mathematically , therefore , the problem can be regarded as a hyperbolic one with stiff relaxation terms which requires special care to capture the dynamics in a stable and accurate manner . in the next section we discuss a simple example of a hyperbolic equation with relaxation which exhibits the problems discussed above and then introduce implicit - explicit ( imex ) runge kutta methods to deal with these kind of equations . in essence , these methods treat the advection character of the system with strong - stability preserving ( ssp ) explicit schemes , while the relaxation character with an l - stable diagonally implicit runge kutta ( dirk ) scheme . after presenting the scheme , its properties and some examples , we discuss in detail its application to the resistive mhd equations . a prototypical hyperbolic equation with relaxation is given by @xmath96 where @xmath97 is the _ relaxation time _ ( not necessarily constant either in space or in time ) , @xmath98 gives rise to a quasilinear system of equations ( _ i.e. _ , @xmath98 depends linearly on first derivatives of @xmath99 ) , and @xmath15 does not contain derivatives of @xmath99 . in the limit @xmath100 ( corresponding for the resistive mhd equations to the case of vanishing conductivity ) the system is hyperbolic with propagation speeds bounded by @xmath101 . this maximum bound , together with the length scale @xmath1 of the system , define a characteristic timescale @xmath102 of the hyperbolic part . in the opposite limit @xmath103 ( corresponding to the case of infinite conductivity ) , the system is instead said to be _ stiff _ , since the timescale @xmath51 of the relaxation ( or stiff ) term @xmath104 is in general much larger than the timescale @xmath105 of the hyperbolic part @xmath98 . in such a limit , the stability of an explicit scheme is only achieved with a timestep size @xmath106 . this requirement is certainly more restrictive than the courant - lewy - friedrichs ( cfl ) stability condition @xmath107 for the hyperbolic part and makes an explicit integration impractical . the development of efficient numerical schemes for such systems is challenging , since in many applications the relaxation time can vary by several orders of magnitude across the computational domain and , more importantly , to much beyond the one determined by the speed @xmath101 . when faced with this issue several strategies can be adopted . the most straightforward one is to consider only the stiff limit @xmath108 , where the system is well approximated by a suitable reduced set of conservation laws called _ `` equilibrium system '' _ @xcite such that @xmath109 where @xmath110 is a reduced set of variables . this approach can be followed if the resulting system is also hyperbolic . this is precisely the case in the resistive mhd equations for vanishing resistivity @xmath111 ( or @xmath112 ) . in this case , the equations reduce to those of ideal mhd and describe indeed an `` equilibrium system '' in which the magnetic field is simply advected with the flow . as discussed earlier , this limit is often adequate to describe the behaviour of dense astrophysical plasmas , but it may also stray away in the magnetospheres . a more general approach could consist of dividing the computational domain in regions in each of which a simplified set of equations can be adopted . as an example , the ideal - mhd equations could be solved in the interior of compact objects , the force - free mhd equations could be solved in the magnetosphere , and finally the maxwell equations for the vacuum regions outside the compact object . however , this approach requires the overall scheme to suitably match the different regions so as to obtain a global solution . this task , unfortunately , is far from being straightforward and , to date , it lacks a rigorous definition . an alternative approach consists of considering the original hyperbolic - relaxation system in the whole computational domain and then employ suitable numerical schemes that work for all regions . among such schemes is the strang - splitting technique @xcite , which has been recently applied by @xcite for the solution of the ( special ) relativistic resistive mhd equations . the strang - splitting scheme provides second - order accuracy if each step is at least second - order accurate , and this property is maintained under suitable assumptions even for stiff problems @xcite . in practice , however , higher - order accuracy is difficult to obtain even in non - stiff regimes with this kind of splitting . moreover , when applied to hyperbolic systems with relaxation , strang - splitting schemes reduce to first - order accuracy since the kernel of the relaxation operator is non - trivial and corresponds to a singular matrix in the linear case , therefore invalidating the assumptions made by @xcite to ensure high - order accuracy . @xcite avoided this problem by solving analytically the stiff part in a reduced form of ampere s law . although this procedure works well for smooth solutions , our implementation of the method has revealed problems when evolving discontinuous flows ( shocks ) for large - conductivities plasmas . moreover , it is unclear whether the same procedure can be adopted in more general configurations , where an analytical solution may not be available . as an alternative approach to the methods solving the relativistic resistive mhd equations on a single computational domain , we here introduce an imex runge - kutta method @xcite to cope with the stiffness problems discussed above . these methods , which are easily implemented , are still under development and have few ( relatively minor ) drawbacks . the most serious one is a degradation to first or second - order accuracy for a range of values of the relaxation time @xmath51 . however , since high - resolution shock - capturing ( hrsc ) schemes usually employed for the solution of the hydrodynamic equations already suffer from similar effects at discontinuities , the possible degradation of the imex schemes does not spoil the overall quality numerical solution when employed in conjunction with hrsc schemes . the next sections review in some detail the imex schemes and our specific implementation for the relativistic resistive mhd equations . the imex runge - kutta schemes rely on the application of an implicit discretization scheme to the stiff terms and of an explicit one to the non - stiff ones . when applied to system ( [ stiff_equation ] ) it takes the form @xcite @xmath113 where @xmath114 are the auxiliary intermediate values of the runge - kutta scheme . the matrices @xmath115 and @xmath116 are @xmath117 matrices such that the resulting scheme is explicit in @xmath118 ( _ i.e. _ , @xmath119 for @xmath120 ) and implicit in @xmath15 . an imex runge - kutta scheme is characterized by these two matrices and the coefficient vectors @xmath121 and @xmath122 . since simplicity and efficiency in solving the implicit part at each step is important , it is natural to consider diagonally implicit runge - kutta ( dirk ) schemes ( _ i.e. _ , @xmath123 for @xmath124 ) for the stiff terms . a particularly convenient way of describing an imex runge - kutta scheme is offered by the butcher notation , in which the scheme is by a double tableau of the type @xcite @xmath125 where the index @xmath9 indicates a transpose and where the coefficients @xmath126 and @xmath127 used for the treatment of non - autonomous systems are given by @xmath128 the accuracy of each of the runge - kutta is achieved by imposing restrictions in some of the coefficients of their respective butcher tableaus . although each of them separately can have an arbitrary accuracy , this does not ensure that the combination of the two schemes will preserve the same accuracy . in addition to the above conditions for each runge - kutta scheme , there are also some additional conditions combining terms in the two tableaus which must be fulfilled in order to achieve a global accuracy order for the complete imex scheme . since the details of these methods are not widely known , we first consider a simple example to fix ideas . a second - order imex scheme can be written in the tableau form given in table [ ssp2 - 222 ] . the intermediate and final steps of this imex runge - kutta scheme would then be written explicitly as @xmath129 \ , , \nonumber \\ & & { \boldsymbol u}^{n+1 } = { \boldsymbol u}^n + \frac{\delta t}{2 } [ f({\boldsymbol u}^{(1 ) } ) + f({\boldsymbol u}^{(2 ) } ) ] \nonumber \\ & & \hskip 1.8 cm + \frac{\delta t}{2 \epsilon } [ r({\boldsymbol u}^{(1 ) } ) + r({\boldsymbol u}^{(2 ) } ) ] \ , . \nonumber\end{aligned}\ ] ] note that at each sub - step an implicit equation for the auxiliary intermediate values @xmath114 must be solved . the complexity of inverting this equation will clearly depend on the particular form of the operator @xmath104 . stable solutions of conservation - type equations are usually analyzed in terms of a suitable norm being bounded in time . with @xmath130 representing the solution vector at the time @xmath131 , then a sequence @xmath132 is said to be _ `` strongly stable '' _ in a given norm @xmath133 provided that @xmath134 for all @xmath135 . c c c c @xmath23 & & @xmath23 & @xmath23 + @xmath136 & & @xmath136 & @xmath23 + & & @xmath137 & @xmath137 + c c c c @xmath138 & & @xmath138 & @xmath23 + @xmath139 & & @xmath140 & @xmath138 + & & @xmath137 & @xmath137 + @xmath141 [ ssp2 - 222 ] the most commonly used norms for analyzing schemes for nonlinear systems are the total - variation ( tv ) norm and the infinity norm . a numerical scheme that maintains strong stability at the discrete level is called strong stability preserving ( ssp ) ( see @xcite for a detailed description of optimal ssp schemes and their properties ) . because of the stability properties of the imex schemes @xcite , it follows that if the explicit part of the imex scheme is ssp , then the method is ssp for the equilibrium system in the stiff limit . this property is essential to avoid spurious oscillations during the evolution of non - smooth data . the stability of the implicit part of the imex scheme is ensured by requiring that the runge - kutta is `` l - stable '' and this represents an essential condition for stiff problems . in practice , this amounts to requiring that the numerical approximation is bounded in cases when the exact solution is bounded . a more strict definition can be derived starting from a linear scalar ordinary differential equation , namely @xmath142 in this case it is easy to define the stability ( or amplification ) function @xmath143 as the ratio of the solutions at subsequent timesteps @xmath144 , where @xmath145 . a runge - kutta scheme is then said to be _ l - stable _ if @xmath146 ( _ i.e. _ , it is bounded ) and @xmath147 @xcite . there are a number of imex runge - kutta schemes available in the literature and we report here only some of the second and third - order schemes which satisfy the condition that in the limit @xmath108 , the solution corresponds to that of the equilibrium system ( [ equilibrium_system ] ) @xcite . these are given in their butcher tableau form in table [ ssp2 - 322 ] and are taken from @xcite . in all these schemes the implicit tableau corresponds to an l - stable scheme . the tableaus are reported in the notation ssp@xmath148 , where @xmath149 denotes the order of the ssp scheme and the triplet @xmath150 characterizes respectively the number of stages of the implicit scheme ( @xmath151 ) , the number of stages of the explicit scheme ( @xmath4 ) , and the order of the imex scheme ( @xmath50 ) . ssp2 @xmath152 0.125 cm c c c c c @xmath153 & & @xmath154 & @xmath154 & @xmath23 + @xmath137 & & @xmath137 & @xmath154 & @xmath23 + @xmath155 & & @xmath137 & @xmath137 & @xmath23 + & & @xmath156 & @xmath156 & @xmath156 0.2 cm c c c c c @xmath157 & & @xmath157 & @xmath154 & @xmath154 + @xmath157 & & @xmath153 & @xmath157 & @xmath154 + @xmath155 & & @xmath156 & @xmath156 & @xmath156 + & & @xmath156 & @xmath156 & @xmath156 + [ ssp2 - 322 ] 0.5 cm ssp3 @xmath152 .125 cm c c c c c @xmath154 & & @xmath153 & @xmath153 & @xmath23 + @xmath158 & & @xmath155 & @xmath153 & @xmath23 + @xmath137 & & @xmath157 & @xmath157 & @xmath23 + & & @xmath159 & @xmath159 & @xmath160 + 0.2 cm c c c c c @xmath138 & & @xmath138 & @xmath23 & @xmath23 + @xmath161 & & @xmath162 & @xmath138 & @xmath23 + @xmath163 & & @xmath164 & @xmath23 & @xmath138 + & & @xmath159 & @xmath159 & @xmath160 + 0.5 cm ssp3 @xmath165 0.125 cm c c c c c c @xmath154 & & @xmath23 & @xmath153 & @xmath153 & @xmath23 + @xmath154 & & @xmath23 & @xmath153 & @xmath153 & @xmath23 + @xmath158 & & @xmath23 & @xmath155 & @xmath153 & @xmath23 + @xmath137 & & @xmath23 & @xmath157 & @xmath157 & @xmath23 + & & @xmath23 & @xmath159 & @xmath159 & @xmath160 + .2 cm c c c c c c @xmath166 & & @xmath166 & @xmath23 & @xmath23 & @xmath23 + @xmath23 & & @xmath167 & @xmath166 & @xmath23 & @xmath23 + @xmath136 & & @xmath23 & @xmath168 & @xmath166 & @xmath23 + @xmath137 & & @xmath169 & @xmath64 & @xmath170 & @xmath166 + & & @xmath23 & @xmath159 & @xmath159 & @xmath160 + @xmath171 having reviewed the main properties of the imex schemes , we now apply them to the particular case of the special relativistic resistive mhd equations . our goal is to consider a numerical implementation of the general system that can deal with standard hydrodynamic issues ( like shocks and discontinuities ) as well as those brought up by the stiff terms discussed in the previous section . hence , we adopt high - resolution shock - capturing algorithms ( see appendix [ appendixb ] ) together with imex schemes . because the first ones involve the introduction of conserved variables in order to cast the equations in a conservative form , we first discuss how to implement the imex scheme within our target system and subsequently how to perform the transformation from the conserved variables to the primitive ones . for our target system of equations it is possible to introduce a natural decomposition of variables in terms of those whose evolution do not involve stiff terms and those which do . more specifically , with the electrical resistivity @xmath64 playing the role of the relaxation parameter @xmath51 , the vector of fields @xmath172 can be split in two subsets @xmath173 , with @xmath174 containing the stiff terms , and @xmath175 the non - stiff ones . following the prototypical eq . ( [ stiff_equation ] ) , the evolution equations for the relativistic resistive mhd equations can then be schematically written as @xmath176 where the relaxation parameter @xmath51 is allowed to depend also on the @xmath177 non - stiff fields . the vector @xmath177 can be evolved straightforwardly as it involves no stiff term . we further note that for our particular set of equations , it is convenient to write the stiff part as @xmath178 as a result , the procedure to compute each stage @xmath114 of the imex scheme can be performed in two steps : 1 . compute the explicit intermediate values @xmath179 from all the previously known levels , that is @xmath180 where we have defined @xmath181 and @xmath182 in eq . ( [ first_stepb ] ) is a simple division and not a contraction on dummy indices . 2 . compute the implicit part , which involves only @xmath183 , by solving @xmath184 note that the implicit equation , with the previous assumption ( [ stiff_part ] ) , can be inverted explicitly @xmath185^{-1 } \ , , \end{aligned}\ ] ] since the form of the matrix @xmath186 $ ] is known explicitly in terms of the evolved fields . the explicit expressions for stiff part are then given simply by @xmath187 with the matrix @xmath188 defined as @xmath189 hence , the matrix @xmath190 can be computed explicitly to obtain @xmath191 where @xmath192 and @xmath193 . + summarizing : first , an intermediate state @xmath194 is found through the evolution of the non - stiff part for the electric field . second , if the velocity @xmath195 is known , the evolution of the stiff part can be performed by acting with @xmath190 to obtain @xmath196\,.\end{aligned}\ ] ] at this point the approach proceeds with the conversion from the conserved variables to the primitive ones . because of the coupling between the electric and the velocity fields , such a procedure is rather involved and more complex than in the ideal - mhd case ; a detailed discussion of how to do this in practice will be presented in sect . [ inversion_con2prim ] . it is interesting to highlight the consistency at two known limits of the implicit solution of the stiff part . in the ideal - mhd limit ( _ i.e. _ , @xmath82 ) the first term of eq . ( [ invert_matrix_e ] ) vanishes , while the contribution of the second term leads to the ideal - mhd condition ( [ ef_imhd ] ) . on the other hand , in the vanishing conductivity limit ( _ i.e. _ , @xmath197 ) the second term in eq . ( [ invert_matrix_e ] ) vanishes , and the matrix reduces to the identity one @xmath198 . in this case , the electric field is obtained only by evolving the explicit part , _ i.e. _ , @xmath199 . finally , it is important to stress that one could , in principle , have considered the alternative route of adopting instead @xmath200 , so that the right - hand - side of @xmath37 would be considered stiff with @xmath201 and @xmath202 . however , this choice could lead to spurious numerical oscillations in the solution since the fluxes of @xmath37 can be discontinuous , while they would be evolved with an implicit runge - kutta . as it has been shown under fairly general conditions , high - order ssp schemes are necessarily explicit @xcite , so it follows that this part of the equations can not be evolved with the implicit runge - kutta unless a low - order scheme is implemented . as mentioned in the previous section , in order to evolve our system of equations , the fluxes @xmath203 must be computed at each timestep . these fluxes depend on the primitive fields @xmath204 , which must be recovered from the evolved conserved fields @xmath205 . these quantities are related by complicated equations which become transcendental except for particularly simple equations of state ( eos ) . as a result , the conversion must be in general pursued numerically and the primitive variables are then given by the roots of the function @xmath206 where @xmath207 is given by the chosen eos and @xmath208 is the trial value for the pressure eventually leading to the primitive variables . note that since @xmath209 [ _ cf . _ eq . ( [ second_step_a ] ) ] , the values of the conserved quantities @xmath210 at time @xmath211 are obtained by evolving their non - stiff evolution equations which , however , provide only an approximate solution for the electric field @xmath212 . as discussed in the previous section , the final solution for the electric field @xmath213 requires the inversion of an implicit equation and , hence , is a function of the velocity @xmath57 and of the fields @xmath214 [ _ cf . _ ( [ invert_matrix_e ] ) ] . however , the velocity is a primitive quantity and thus not known at the time @xmath211 . it is clear , therefore , that it is necessary to obtain , at the same time , the evolution of the stiff part of the equations and the conversion of the conserved quantities into to the primitive ones . in what follows we describe how to do this in practice using an iterative procedure . 1 . adopt as initial guess for the velocity its value at the previous time level @xmath215 . the electric field @xmath213 is computed by eq . ( [ invert_matrix_e ] ) as a function of @xmath216 . 2 . adopt as initial guess for the pressure its value at the previous time level @xmath217 . compute in the following order + @xmath218 3 . solve numerically eq . ( [ trascendental ] ) by means of an iterative newton - raphson solver , so that the solution at the iteration @xmath219 can be computed as @xmath220 the derivative of the function @xmath221 needed for the newton - raphson solver can be computed as @xmath222 with @xmath223 being the local speed of the fluid which , for an ideal - fluid eos @xmath224 is given by @xmath225 4 . with the newly obtained values for the velocity @xmath57 and the pressure @xmath50 , the steps ( i)(iii ) can be iterated until the difference between two successive values falls below a specified tolerance . the approach discussed above is a simple procedure that can be implemented straightforwardly and works well for moderate ratios of @xmath226 , converging in less than @xmath227 iterations both for smooth electromagnetic fields and for discontinuous ones . faster and more robust procedures to obtain the primitive variables certainly ca be implemented , but this is beyond the scope of this work . in this section we present several one - dimensional ( 1d ) or two - dimensional ( 2d ) tests which have been used to validate the implementation of the imex runge - kutta schemes in the different regimes of relativistic resistive mhd . in all these tests we employ the ideal - fluid eos with @xmath228 for the 1d tests and @xmath229 in the 2d ones . the different tests span several prescriptions for the conductivity and compare the solutions obtained either with those expected in the ideal - mhd limit or with those obtained with the strang - splitting technique . more specifically , in 1d we consider large - amplitude circularly polarized ( cp ) alfvn waves to test the ability of the code to reproduce the ideal - mhd results when adopting a very large conductivity . the intermediate conductivity regime is instead tested by simulating a self - similar current sheet . finally , a large range of uniform and non - uniform conductivities are used for a representative shock - tube problem . in 2d , on the other hand , we first consider a commonly employed test for ideal - mhd codes corresponding to a cylindrical explosion . subsequently , we simulate a toy model for a `` magnetized neutron star '' when modelled as a cylindrically symmetric density distribution obeying a gaussian - profile . the behaviour of the magnetic field is studied again for a range of constant and non - uniform conductivities . this test is discussed in detail by @xcite and we report here only a short summary . the solution describes the propagation of a large amplitude circularly - polarized alfvn waves along a uniform background field @xmath230 in a domain with periodic boundary conditions . the exact solution in the ideal - mhd limit and assuming @xmath231 for simplicity , is given by @xcite @xmath232 , \sin[k(x - v_a~t)])\ , , \nonumber \\ ( v_y , v_z ) & = & -\frac{v_a}{b_0 } ~(b_y , b_z ) \,,\end{aligned}\ ] ] where @xmath233 , @xmath149 is the wave vector , @xmath234 is the amplitude of the wave and the special relativistic alfvn speed @xmath235 is given by @xmath236 in practice , using such ideal - mhd solution it is possible to assess the accuracy of evolution of the resistive equations by requiring that for very large conductivities the numerical solution approaches the exact one as the resolution is progressively increased . it is also worth remarking that although we do not expect the solution of the resistive mhd equations to converge to that of ideal mhd for any finite value of @xmath4 , we also expect the differences between the two to be @xmath237 and thus negligibly small for sufficiently large values . for this reason , we have performed the evolution with a high uniform conductivity of @xmath238 for three different resolutions @xmath239 covering the computational domain @xmath240 $ ] . in addition , the initial data parameters have been chosen so that @xmath241 and @xmath242 , thus yielding @xmath243 , with a full period being achieved at @xmath244 . 1.0 cm fig . [ alfven ] confirms this expectation by reporting the component @xmath245 after one period and thus overlapping with the initial one ( at @xmath246 ) for the highest resolution . this test shows clearly that in the limit of very high conductivity the resistive mhd equations tend to a solution which is very close to the same solution obtained in the ideal - mhd limit . the convergence rate measured for the different fields is consistent with the second - order spatial discretization being used as expected for smooth flows ( see appendix [ appendixb ] ) . the details of this test are described by @xcite , so again we provide here only a short description for completeness . we assume that the magnetic pressure is much smaller than the fluid pressure everywhere , with a magnetic field given by @xmath247 , where @xmath248 changes sign within a thin current layer of width @xmath249 . provided the initial solution is in equilibrium ( @xmath250 ) , the evolution is a slow diffusive expansion of the layer due to the resistivity and described by the diffusion equation [ _ cf . _ eq . ( [ newtonian_rmhd ] ) with @xmath251 @xmath252 as the system expands , the width of the layer becomes much larger than @xmath249 and it evolves in a self - similar fashion . for @xmath253 , the analytical exact solution is given by @xmath254 where @xmath255 and `` @xmath256 '' is the error function . this solution can be used for testing the moderate resistive regime . following @xcite , and in order to avoid the singular behaviour at @xmath246 , we have chosen as initial data the solution at @xmath257 with @xmath258 , @xmath259 , @xmath260 and @xmath261 . the domain covers the region @xmath262 $ ] with @xmath263 points . 1.0 cm the numerical simulation is evolved up to @xmath264 and then the numerical and the exact solution are compared in fig . [ sheet ] . the two solutions match so well that they are not distinguishable on the plot , thus , showing that the intermediate - conductivity regime is also well described by our method . as prototypical shock - tube test we consider a simple mhd version of the brio and wu test @xcite , where the initial left and right states are separated at @xmath265 and are given by @xmath266 while all the other fields set to @xmath23 . we consider both uniform and non - uniform conductivities . in the latter case we adopt the following prescription @xmath267 thus allowing for nonlinearities in the dependence of the conductivity on the conserved quantity @xmath268 . this is one of the simplest cases , but in realistic situations a more general expression for the conductivity can be assumed , where @xmath4 is a function of both the rest - mass density and of the specific internal energy , _ i.e. _ , @xmath269 . the exact solution of the ideal mhd riemann problem was found by @xcite , and in our particular case it has been computed with a publicly available code [ see @xcite ] . when @xmath270 , the structure of the solution contains only two fast waves , a rarefaction moving to the left and a shock moving to the right , with a tangential discontinuity between them . more demanding riemann problems have also been performed but the procedure to convert the conserved variables into the primitive ones has shown in these case a lack of robustness for large ratios of @xmath226 . we have first considered the case of uniform ( @xmath271 ) and very large conductivity ( @xmath272 ) as in this case we can use the solution in the ideal - mhd limit as a useful guide . the profile of the magnetic field component @xmath245 for three different resolutions @xmath273 and the exact solution are shown in the left panel of fig . [ shock_convergence ] at @xmath274 . overall , the results indicate that even in the presence of shocks our numerical solution of the resistive mhd tends to the ideal - mhd solution as the resolution is increased . it is also interesting to study the behaviour of the solution for different values of the constant @xmath275 while still keeping a uniform conductivity ( _ i.e. _ , @xmath271 ) . this is shown in the right panel of fig.[shock_convergence ] , which displays the different solutions obtained , and where it is possible to see how they change smoothly from a wave - like solution for @xmath276 to the ideal - mhd one for @xmath272 . this set up is also useful to perform a comparison between the imex and the strang - splitting approaches . in fig . [ error ] we show the @xmath277-norm of the difference between the numerical solution obtained with both schemes and the ideal - mhd exact solution , for different values of the conductivity with @xmath278 points . several comments are in order . firstly , the reported difference between the numerical solution for the resistive mhd equations and the ideal - mhd equations should not be interpreted as an error given that the latter is not the correct solution of the equations . hence , the fact that the use of a strang - splitting method yields smaller differences is simply a measure of its ability of better capture steep gradients . secondly , while the imex approach does not show any sign of instability for @xmath275 ranging between @xmath279 and @xmath280 , the implementation adopting the strang - splitting technique becomes unstable for moderately high values of the conductivity and , at least for the shock - tube problem , no numerical solution was possible for @xmath281 at the above resolution . increasing the resolution can help increase the maximum value of the resistivity which can be handled , but since this gain is only linear with the number of gridpoints aiming for higher conductivities results impractical . this is shown in the inset of fig . [ error ] , which reports the maximum conductivity for which a solution was possible , @xmath282 , as a function of the number of gridpoints , @xmath283 . finally , we note that the difference between the imex numerical solution and the exact ideal - mhd one saturates between @xmath284 . this is not surprising since the differences are expected to be @xmath285 , and thus the saturation in the differences essentially provides a measure of our truncation error at the resolution used . a more challenging test is offered by the solution of the shock - tube in the presence of a non - uniform conductivity . in particular , we have considered the same initial states and the same non - uniform conductivity discussed above , but used different values for the exponent @xmath138 in ( [ def_cond ] ) while keeping @xmath275 constant . the results of this test are shown in the left panel of fig . [ shock_condpot ] , where the conductivity is plotted at @xmath274 for several values of @xmath138 . note that the conductivity traces the evolution of the rest - mass density and that the solution can be found also when @xmath4 varies of almost @xmath286 orders of magnitude across the grid . similarly , the right panel of fig . [ shock_condpot ] displays the component @xmath245 for the different values of @xmath138 . it should be stressed that because of the relation ( [ def_cond ] ) between @xmath4 and @xmath10 , the region on the left has at this time a very high conductivity and the numerical solution tends to the ideal - mhd one . the opposite happens on the right region , where the conductivity is lower for higher values of @xmath138 . clearly , the results presented in fig . [ shock_condpot ] show that our implementation can handle non - uniform ( and quite steep ) conductivity profiles even in the presence of shocks . 1.cm we now consider problems involving shocks in more than one dimension . a demanding test for the relativistic codes is the cylindrical blast wave expanding in a plasma with an initially uniform magnetic field . although there is no exact solution for this problem , strong symmetric explosions are useful tests since shocks are present in all the possible directions and the numerical implementation is therefore tested in all of its parts . for this test we set a square domain @xmath287 $ ] with a resolution @xmath288 . the initial data is such that inside the radius @xmath289 the pressure is set to @xmath290 while the density to @xmath291 . in the intermediate region @xmath292 the two quantities decrease exponentially up to the exterior region @xmath293 , where the ambient fluid has @xmath294 . the magnetic field is uniform with only one nontrivial component @xmath295 . the other fields are set to be zero ( _ i.e. _ , @xmath296 ) , which is consistent within the ideal - mhd approximation . the evolution is performed with a high conductivity @xmath297 in order to recover the solution from the ideal - mhd approximation . as shown in fig . [ explosion_bs_2d ] , which reports the magnetic field components @xmath298 ( left panel ) and @xmath245 ( right panel ) at time @xmath299 , we obtain results that are qualitatively similar to those published in different works @xcite . while a strict comparison with an exact solution is not possible in this case , the solution found matches extremely well the one obtained with another 2d code solving the ideal mhd equations . most importantly , however , the figure shows that the solution is regular everywhere and that similar results can be obtained also with smaller values of the conductivity ( _ e.g. _ , no significant difference was seen for @xmath300 ) . we next consider a toy model for a star , thought as an infinite column of fluid aligned with the @xmath301-axis but with compact support in other directions . because of the symmetry in the @xmath301-direction , @xmath302 for all the fields and the problem is therefore two - dimensional . more specifically , we consider initial data given by @xmath303 where @xmath304 is the cylindrical radial coordinate . the other fields can be computed at the initial time by using the polytropic eos @xmath305 , the ideal - mhd expression ( [ ef_imhd ] ) for the electric field , and the electric charge from the constraint equation @xmath306 . we have chosen @xmath307 , @xmath308 , @xmath309 and @xmath310 . an atmosphere ambient fluid with @xmath291 is added outside the cylinder . finally , the resolution is @xmath311 and the domain is @xmath312 $ ] . this simple problem exhibits some of the issues present in a magnetized rotating neutron star : a compactly supported rest - mass density distribution , an azimuthal velocity field and a poloidal magnetic field . suitable source terms describing a gravitational potential have been added to the euler equations in order to get , at least at the initial time , a stationary solution . in the ideal - mhd limit the magnetic lines are frozen in the fluid and thus a static profile is also expected for the magnetic field . in the left panel of fig . [ star_bz ] we plot the slice @xmath313 of the magnetic field component @xmath314 at @xmath315 as obtained from the evolution of the resistive mhd system for different uniform conductivities in the range @xmath316 $ ] . in the limiting case @xmath317 the solution corresponds to a wave propagating at the speed of light ( _ i.e. _ , the solution of the maxwell equations in vacuum ) , while for large values of @xmath275 the solution is stationary ( as expected in the ideal - mhd limit ) . the behaviour observed in the left panel fig . [ star_bz ] is also the expected one : the higher the conductivity , the closer the solution is to the stationary solution of the ideal - mhd limit . for low conductivities , on the other hand , there is a significant diffusion of the solution , which is quite rapid for @xmath318 and for this reason those values are not plotted here . we note that values of the conductivity larger than @xmath319 lead to numerical instabilities that we believe are coming from inaccuracies in the evolution of the charge density @xmath37 , and which contains spatial derivatives of the current vector . in addition , the stiff quantity @xmath320 is seen to converge only to an order @xmath321 . this can be due to the `` final layer '' problem of the imex methods , which is known to produce a degradation on the accuracy of the stiff quantities . luckily , this does not spoil the convergence of the non - stiff fields , which are instead second - order convergent . it is possible that the use of stiffly - accurate schemes can solve this degradation of the convergence and this is an issue we are presently exploring . 1.0 cm we finally consider the same test , but now employing the non - uniform conductivity given by eq . ( [ def_cond ] ) with @xmath272 and different values for @xmath138 . the results are presented in the right panel of fig . [ star_bz ] , which shows that the magnetic fields inside the star are basically the same in all the cases , stressing the fact that the interior of the star will not be significantly affected by the exterior solution , which has much smaller conductivity . however , the electromagnetic fields outside the star do change significantly for different values of @xmath138 , underlining the importance of a proper treatment of the resistive effects in those regions of the plasma where the ideal - mhd approximation is not a good one . we have introduced implicit - explicit runge - kutta schemes to solve numerically the ( special ) relativistic resistive mhd equations and thus deal , in an effective and robust way , with the problems inherent to the evolution of stiff hyperbolic equations with relaxation terms . since for these methods the only limitation on the size of the timestep is set by the standard cfl condition , the approach suggested here allows to solve the full system of resistive mhd equations efficiently without resorting to the commonly adopted limit of the ideal - mhd approximation . more specifically , we have shown that it is possible to split the system of relativistic resistive mhd equations into a set of equations that involves only non - stiff terms , which can be evolved straightforwardly , and a set involving stiff terms , which can also be solved explicitly because of the simple form of the stiff terms . overall , the only major difficulty we have encountered in solving the resistive mhd equations with imex methods arises in the conversion from the conserved variables to the primitive ones . in this case , in fact , there is an extra difficulty given by the fact that there are four primitive fields which are unknown and have to be inverted simultaneously . we have solved this problem by using extra iterations in our 1d newton - raphson solver , but a multidimensional solver is necessary for a more robust and efficient implementation of the inversion process . with this numerical implementation we have carried out a number of numerical tests aimed at assessing the robustness and accuracy of the approach , also when compared to other equivalents ones , such as the strang - splitting method recently proposed by @xcite . all of the tests performed have shown the effectiveness of our approach in solving the relativistic resistive mhd equations in situations involving both small and large uniform conductivities , as well as conductivities that are allowed to vary nonlinearly across the plasma . furthermore , when compared with the strang - splitting technique , the imex approach has not shown any of the instability problems that affect the strang - splitting approach for flows with discontinuities and large conductivities . while the results presented here open promising perspectives for the implementation of imex schemes in the modelling of relativistic compact objects , at least two further improvements can be made with minor efforts . the first one consists of the generalization of the ( special ) relativistic resistive mhd equations with a scalar isotropic ohm s law to the general relativistic case , and its application to compact astrophysical bodies such a magnetized binary neutron stars @xcite . the solution of the resistive mhd equations can yield different results not only in the dynamics of the magnetosphere produced after the merger , but also provide the possibility to predict , at least in some approximation , the electromagnetic radiation produced by the merger of these objects . the second improvement consists of considering a non - scalar and anisotropic ohm s law , so that the behaviour of the currents in the magnetosphere can be described by using a very high conductivity along the magnetic lines and a negligibly small one in the transverse directions @xcite . such an improvement may serve as a first step towards an alternative modelling of force - free plasmas . we would like to thank eric hirschmann , serguei komissarov , steve liebling , jonathan mckinney , david neilsen and olindo zanotti for useful comments and bruno giacomazzo for comments and for providing the code computing the exact solution of the riemann problem in ideal mhd . ll and cp would like to thank famaf ( unc ) for hospitality . cp is also grateful to lorenzo pareschi for the many clarifications about the imex schemes . this work was supported in part by nsf grants phy-0326311 , phy-0653369 and phy-0653375 to louisiana state university , the dfg grant sfb / transregio 7 , conicet and secyt - unc . we are generically interested in solving hyperbolic conservation laws of the form @xmath322 where @xmath172 is the vector of the evolved fields , @xmath323 are their fluxes and @xmath324 contains the sources terms . the semi - discrete version of this equation , in one dimension , is simply given by @xmath325 where @xmath326 are consistent numerical fluxes evaluated at the interfaces between numerical cells . these consistent fluxes are computed by using hrsc methods , which are based on the use of riemann solvers . more specifically , we have implemented a modification of the local lax - friedrichs approximate riemann solver introduced by @xcite , which only needs the spectral radius ( _ i.e. _ , the maximum eigenvalue ) of the system . in highly relativistic cases , like the ones we are interested in , the spectral radius is close to the light speed @xmath21 and so the local lax - friedrichs reduces to the simpler lax - friedrichs flux @xmath327\,,\end{aligned}\ ] ] where @xmath328 are the reconstructed solutions on the left and on the right of the interface and @xmath329 their corresponding fluxes . the standard procedure is then to reconstruct the solution @xmath328 by interpolating with a polynomial and then compute the fluxes @xmath330 and @xmath331 . in our implementation we first recombine the fluxes and the solution as @xcite @xmath332 then , using a piecewise linear reconstruction , these combinations can be computed on the left / right of the interface as @xmath333 where @xmath334 are just the slopes used to extrapolate @xmath335 to the interfaces . finally , the consistent flux is computed by a simple average @xmath336\,.\end{aligned}\ ] ] for a linear reconstruction the slopes can be written as @xmath337 so that it is trivial to check that the standard lax - friedrichs ( [ lax - friedrichs ] ) is recovered when @xmath338 . the choice of these slopes becomes crucial in the presence of shocks or very sharp profiles , while the use of some nonlinear operators @xmath339 preserves the total variation diminishing ( tvd ) condition on the interpolating polynomial . in this way , the tvd schemes capture accurately the dynamics of strong shocks without the oscillations which appear with standard finite - difference discretizations . monotonicity is typically enforced by making use of slope limiters and we have in particular implemented the monotonized centered ( mc ) limiter @xmath340 ~{\rm min}(2\|x\|,2\|y\|,\frac{1}{2}\|x+y\| ) \,,\ ] ] which provides a good compromise between robustness and accuracy . note that with linear reconstruction the scheme is second - order accurate in the smooth regions , although it drops to first order near shocks and at local extrema .	many astrophysical processes involving magnetic fields and quasi - stationary processes are well described when assuming the fluid as a perfect conductor . for these systems , the ideal - magnetohydrodynamics ( mhd ) description captures the dynamics effectively and a number of well - tested techniques exist for its numerical solution . yet , there are several astrophysical processes involving magnetic fields which are highly dynamical and for which resistive effects can play an important role . the numerical modeling of such non - ideal mhd flows is significantly more challenging as the resistivity is expected to change of several orders of magnitude across the flow and the equations are then either of hyperbolic - parabolic nature or hyperbolic with stiff terms . we here present a novel approach for the solution of these relativistic resistive mhd equations exploiting the properties of implicit - explicit ( imex ) runge kutta methods . by examining a number of tests we illustrate the accuracy of our approach under a variety of conditions and highlight its robustness when compared with alternative methods , such as the strang - splitting . most importantly , we show that our approach allows one to treat , within a unified framework , both those regions of the flow which are fluid - pressure dominated ( such as in the interior of compact objects ) and those which are instead magnetic - pressure dominated ( such as in their magnetospheres ) . in view of this , the approach presented here could find a number of applications and serve as a first step towards a more realistic modeling of relativistic astrophysical plasmas . [ firstpage ] relativity mhd plasmas methods : numerical
given an arbitrary real sequence @xmath1 , a continued fraction expansion of the form @xmath2 is called a @xmath0-fraction if the parameters @xmath3 $ ] , @xmath4 . it terminates and equals a rational function if @xmath5 for some @xmath4 . if @xmath6 , @xmath4 , the @xmath0-fraction still converges uniformly on every compact subsets of the slit domain @xmath7 ( see ( * ? ? ? * theorem 27.5 ) and ( * ? ? ? * corollary 4.60 ) ) , and in this case , will represent an analytic function , say @xmath8 . such @xmath0-fractions are found having applications in diverse areas like number theory @xcite , dynamical systems @xcite , moment problems and analytic function theory @xcite . in particular , ( * ? ? ? * theorem 69.2 ) , the hausdorff moment problem @xmath9 has a solution if and only if corresponds to a power series of the form @xmath10 , @xmath11 . further , the @xmath0-fractions have also been used to study the geometric properties of ratios of gaussian hypergeometric functions as well as their @xmath12-analogues , ( see the proofs of ( * ? ? ? * theorem 1.5 ) and ( * ? ? ? * theorem 2.2 ) ) . among several such results , one of the most fundamental result concerning @xmath0-fraction is ( * ? ? ? * theorem 74.1 ) in which holomorphic functions having positive real part in @xmath13 are characterised . precisely , @xmath14 is positive if and only if @xmath8 has a continued fraction expansion of the form . moreover , @xmath8 has the integral representation @xmath15,\end{aligned}\ ] ] where @xmath16 is a bounded non - decreasing function having a total increase 1 . many interesting results are also available in literature if we consider subsets of @xmath7 . for instance , let @xmath17 be the class of holomorphic functions having a positive real part on the unit disk @xmath18 . such functions denoted by @xmath19 are called carathodory functions and have the riesz - herglotz representation ( * ? ? ? * theorem 73.1 ) @xmath20 where @xmath12=@xmath21 . further , if @xmath22 is such that @xmath23 and normalised by @xmath24 , then the following continued fraction expansion can be derived @xcite @xmath25 where @xmath26 . note that here @xmath27 . closely related to the carathodory functions are the schur functions @xmath28 given by @xmath29 from , it is clear that @xmath28 maps the unit disk @xmath30 to the closed unit disk @xmath31 . in fact , if @xmath32 describes a one - one correspondence between the classes of holomorphic functions @xmath33 and @xmath17 . the class @xmath33 studied by j. schur @xcite , is the well - known schur algorithm . this algorithm generates a sequence of rational functions @xmath34 from a given sequence @xmath35 of complex numbers lying in @xmath31 . then , with @xmath36 , @xmath37 , satisfying some positivity conditions , @xmath38 , @xmath39 , where @xmath40 is a schur function @xcite . it is interesting to note that @xmath41 , @xmath37 where @xmath42 . moreover , using these parameters @xmath43 , the following schur fraction can be obtained @xcite @xmath44 where @xmath45 is related to the @xmath46 occurring in by @xmath47 , @xmath48 . similar to @xmath0-fractions the schur fraction also terminates if @xmath49 for some @xmath50 . it may be noted that such a case occurs if and only if @xmath28 is a finite blashke product @xcite . let @xmath51 and @xmath52 denote the @xmath53 partial numerator and denominator of respectively . with the initial values @xmath54 , @xmath55 , @xmath56 and @xmath57 , the following recurrence relations hold @xcite @xmath58 @xmath59 using in , we get @xmath60 the relations and are sometimes written in the more precise matrix form as @xmath61 it is also known that @xcite @xmath62 here , and in what follows , @xmath63 for any polynomial @xmath64 with complex coefficients and of degree @xmath65 . from , it follows that @xmath66 further , the even approximants of the schur fraction coincide with the the @xmath53 approximant of the schur algorithm , so that @xmath67 converges to the schur function @xmath28 as @xmath39 . from the point of view of their applications , it is obvious that the parameters @xmath68 of the @xmath0-fraction contain hidden information about the properties of the dynamical systems or the special functions they represent . one way to explore this hidden information is through perturbation ; that is , through a study of the consequences when some disturbance is introduced in the parameter sequence @xmath69 . the main objective of the present manuscript is to study the structural and qualitative aspects of two perturbations . the first is when a finite number of parameters @xmath46 s are missing in which case we call the corresponding @xmath0-fraction as a _ the second case is replacing @xmath70 by a new sequence @xmath71 in which the @xmath72 term @xmath46 is replaced by @xmath73 . the first case is illustrated using gaussian hypergeometric functions , where we use the fact that many @xmath0-fractions converge to ratios of gaussian hypergeometric functions in slit complex domains . the second case is studied by applying the technique of coefficient stripping @xcite to the sequence of schur parameters @xmath74 . this follows from the fact that the schur fraction and the @xmath0-fraction are completely determined by the related schur parameters @xmath75 s and the @xmath0-parameters respectively , and that a perturbation in @xmath45 produces a unique change in the @xmath46 and vice - versa . the manuscript is organised as follows . section [ sec : gap g fractions and structural relations ] provides structural relations for the three different cases provided by the gap @xmath0-fraction . a particular ratio of gaussian hypergeometric functions is used to illustrate the results . the modified @xmath0-fractions given in section [ sec : gap g fractions and structural relations ] has a shift in @xmath76 to @xmath77 and so on for any fixed @xmath78 . instead , the effect of changing @xmath76 to any another value @xmath79 is discussed in section [ sec : perturbed schur parameters ] . illustrations of the results obtained in section [ sec : gap g fractions and structural relations ] leading to characterization of a class of ratio of hypergeometric functions such as pick functions is outlined in section [ sec : a class of pick functions and schur functions ] . as the name suggests , gap-@xmath0-fractions correspond to the @xmath0-sequence @xmath1 with missing parameters . we study three cases in this section and in each the concept of tail sequences of a continued fraction plays an important role . for more information on the tails of a continued fraction , we refer to ( * ? ? ? * chapter ii ) . for @xmath11 , let @xmath8 be the continued fraction and @xmath80 note that is obtained from by removing @xmath76 for some arbitrary @xmath78 which can not be obtained by letting @xmath81 . let , @xmath82 so that @xmath83 is the @xmath84 tail of @xmath8 . we note that ( * ? ? ? * theorem 1 , p.56 ) , the existence of @xmath8 guarantees the existence of @xmath85 . further , if @xmath86 then , from and we obtain the rational function @xmath87 it is known ( * ? ? ? * theorem 2.1 ) , that the @xmath88 approximant of is given by the rational function @xmath89 and that @xmath90 then , @xmath91 } { \mathcal{y}_{k}(0;z)[\mathcal{y}_{k}(0;z)-h(k;z)\mathcal{y}_{k-1}(0;z)]}\\ & = \dfrac{h(k;z)z^{k-1}\prod_{j=1}^{k-1}(1-g_{j-1})g_j } { \mathcal{y}_{k}(0;z)[\mathcal{y}_{k}(0;z)-h(k;z)\mathcal{y}_{k-1}(0;z)]},\end{aligned}\ ] ] where the last equality follows from ( * ? ? ? * eqn.(2.1.9 ) ) . denoting @xmath92 , @xmath48 , we have from @xmath93 ^ 2}.\end{aligned}\ ] ] in the sequel , by @xmath8 we will mean the unperturbed @xmath0-fraction as given in with @xmath94 $ ] , @xmath95 . further , as the notation suggests , the rational function @xmath96 is independent of the parameter @xmath76 and is known whenever @xmath8 is given . the information of the missing parameter @xmath76 at the @xmath88 position is stored in @xmath97 and hence the notation @xmath98 . it may also be noted that the polynomials @xmath99 can be easily computed from the wallis recurrence ( * ? ? ? ( 2.1.6 ) ) @xmath100 with the initial values @xmath101 . thus , we state our first result . [ thm : structural relation for gk missing ] suppose @xmath8 is given . let @xmath98 denote the perturbed @xmath0-fraction in which the parameter @xmath76 is missing . then , with @xmath92 , @xmath48 @xmath102 ^ 2},\end{aligned}\ ] ] where @xmath103 , @xmath96 and @xmath104 are respectively , the @xmath88 partial denominator , the @xmath88 approximant and the @xmath84 tail of @xmath8 . it may be observed that the right side of is of the form @xmath105 with @xmath106 being well defined polynomials . rational functions of such form are said to be rational transformation of @xmath97 and occur frequently in perturbation theory of orthogonal polynomials . for example , see @xcite . a similar result for the perturbed @xmath0-fraction in which a finite number of consecutive parameters are missing can be obtained by an analogous argument that is stated directly as [ thm : structural relation for finite number of parameters missing ] let @xmath8 be given . let @xmath107 denote the perturbed @xmath0-fraction in which the l consecutive parameters @xmath108 are missing . then , @xmath109 ^ 2},\end{aligned}\ ] ] where @xmath110 is the @xmath111 tail of @xmath8 . the next result is about the perturbation in which only two parameters @xmath76 and @xmath112 are missing , where @xmath113 need not be @xmath114 . [ thm : structural relation for gk and gl missing ] let @xmath8 be given . let @xmath115 denote the perturbed @xmath0-fraction in which two parameters @xmath76 and @xmath112 are missing , where we assume @xmath116 , @xmath117 . then @xmath118 ^ 2}\end{aligned}\ ] ] where @xmath119 is the perturbed @xmath84 tail of @xmath8 in which @xmath112 is missing and is given by @xmath120 ^ 2- \mathcal{y}^{(k+1)}_{m-1}(0;z)\mathcal{y}^{(k+1)}_{m}(0;z)h(k+m+1;z)}\end{aligned}\ ] ] where @xmath121 and @xmath122 are respectively , the @xmath123 partial denominator and @xmath123 approximant of the @xmath84 tail of @xmath8 . here , @xmath124 is the @xmath125 tail of @xmath8 . let @xmath126 so that @xmath127 is the perturbed @xmath84 tail of @xmath8 in which @xmath112 is missing . then we can write @xmath128 where @xmath129 . now , proceeding as in theorem @xmath130 , we obtain @xmath131 ^ 2}\end{aligned}\ ] ] hence all that remains is to find the expression for @xmath132 or @xmath133 . now , let @xmath134 and @xmath135 . from and ( * eqn.(1.1.4 ) , p.57 ) , we have @xmath136 it is clear that , the rational function @xmath137 $ ] is the approximant of the @xmath84 tail @xmath83 of @xmath8 . then , using ( * ? ? ? ( 1.1.6 ) , p. 57 ) we obtain @xmath138 } { \mathcal{y}^{(k+1)}_{l - k-1}(0;z)[\mathcal{y}^{(k+1)}_{l - k-1}(0;z)-h(l;z)\mathcal{y}^{(k+1)}_{l - k-2}(0;z)]}\end{aligned}\ ] ] finally , using the fact that @xmath116 , we obtain @xmath139 ^ 2-\mathcal{y}^{(k+1)}_{k+m}(0;z)\mathcal{y}^{(k+1)}_{k+m+1}(0;z)h(k+m+1;z)},\end{aligned}\ ] ] where @xmath140 is the @xmath123 approximant of the @xmath84 tail of @xmath8 . as mentioned earlier , from , and , it is clear that tail sequences play a significant role in deriving the structural relations for the gap @xmath0-fractions . we now illustrate the role of tail sequences using particular @xmath0-fraction expansions . the gaussian hypergeometric function , with the complex parameters @xmath141 , @xmath142 and @xmath143 is defined by the power series @xmath144 where @xmath145 and @xmath146 is the pochhammer symbol . two hypergeometric functions @xmath147 and @xmath148 are said to be contiguous if the difference between the corresponding parameters is at most unity . a linear combination of two contiguous hypergeometric functions is again a hypergeometric function . such relations are called contiguous relations and have been used to explore many hidden properties of the hypergeometric functions ; for example , many special functions can be represented by ratios of gaussian hypergeometric functions . for more details , we refer to @xcite . consider the gauss continued fraction @xcite ( with @xmath149 and @xmath150 ) @xmath151 where @xmath152 let @xmath153 , @xmath154 . we aim to find the ratio of hypergeometric functions given by the continued fraction @xmath155 for this , first note that from , we can write @xmath156= 1-\dfrac{(1-k_1)\omega}{1-\dfrac{k_1(1-k_2)\omega}{1-\dfrac{k_2(1-k_3)\omega } { 1-\cdots}}}\end{aligned}\ ] ] now , replacing @xmath149 and @xmath150 in the contiguous relation ( * ? ? ? 89.6 , p. 336 ) we obtain @xmath157 hence , with @xmath158 , we have @xmath159\\ & = 1-\dfrac{c - b}{c}\dfrac{f(a+1,b;c+1;\omega)}{f(a , b;c;\omega)}\\ & = ( 1-\omega)\dfrac{f(a+1,b;c;\omega)}{f(a , b;c;\omega)},\end{aligned}\ ] ] where the last equality follows from the relation @xmath160 which is easily proved by comparing the coefficients of @xmath161 on both sides . finally , using the well known result ( * ? ? ? 75.3 , p. 281 ) , we obtain @xmath162 note that the continued fraction has also been derived by different means in @xcite and studied in the context of geometric properties of hypergeometric functions . for further analysis , we establish the following formal continued fraction expansion : @xmath163 suppose for now , the left hand side of is denoted by @xmath164 . again using , gauss continued fraction ( * ? ? ? * eqn . 89.9 , p. 337 ) with @xmath165 and @xmath166 , we arrive at @xmath167^{-1}(\omega)=1-\dfrac{b(c - a)}{c(c+1)}\omega \dfrac{f(a+1,b+1;c+2;\omega)}{f(a+1,b;c+1;\omega)}.\end{aligned}\ ] ] in the relation , interchanging @xmath141 and @xmath142 and then replacing @xmath165 and @xmath166 , we get @xmath168 which implies @xmath169 as mentioned , the right hand side is only a formal expansion for the left hand side in . however , note the fact that the sequence @xmath170 , @xmath171 satisfies the difference equation @xmath172 where @xmath173 thus , using ( * ? ? ? * theorem 1 , pp . 295 - 296 ) , we can conclude that the right side of indeed corresponds as well as converges to the left side of , which we state as [ prop : correspondence and convergence of g1 ] the following correspondence and convergence properties hold . a. with @xmath141 , @xmath142 , @xmath145 complex constants , @xmath174 b. the continued fraction on the right side of converges to the meromorphic function @xmath175 in the cut - plane @xmath176 where @xmath177 and @xmath178 . the convergence is uniform on every compact subset of @xmath179 . we would like to mention here that the polynomial sequence @xmath180 corresponding to @xmath181 and that arises during the discussion of the convergence of gauss continued fractions @xcite is given by @xmath182 also note that the continued fraction used in the right side of is @xmath183 here , we recall that @xmath184 , @xmath37 , where @xmath69 are the parameters appearing in the gauss continued fraction . the following result gives a kind of generalization of proposition @xmath185 . the correspondence and convergence properties of the continued fractions involved can be discussed similar to the one for @xmath164 . [ prop : continued fraction identity for gk ] let , @xmath186 then , @xmath187 the case @xmath188 has already been established in proposition @xmath185 . comparing the continued fractions for @xmath189 and @xmath190 , @xmath48 , it can be seen that @xmath189 can be obtained for @xmath190 @xmath48 by shifting @xmath165 , @xmath191 and @xmath192 . for @xmath193 , @xmath48 , we note that the continued fraction in right side of is nothing but the gauss continued fraction ( * ? ? ? 89.9 , p. 337 ) with the shifts @xmath194 , @xmath195 and @xmath196 in the parameters . instead of starting with @xmath197 , as the first partial numerator term in the continued fraction , a modification by inserting a new term changes the hypergeometric ratio given in proposition [ prop : continued fraction identity for gk ] , thus leading to interesting consequences . we state this result as follows . [ thm : gap g fraction identity with minimal parameters ] let @xmath198 then , @xmath199 denoting , @xmath200 we find from ( * ? ? ? 75.3 , p. 281 ) , @xmath201 hence , we need to derive the functions @xmath202 . for @xmath193 , @xmath48 , using and @xmath203 , we find that @xmath204= \dfrac{c - b+j}{c+2j}\omega\dfrac{f(a+j+1,b+j;c+2j+1;\omega ) } { f(a+j , b+j;c+2j;\omega)}\end{aligned}\ ] ] shifting @xmath194 , @xmath205 and @xmath206 in , we find that @xmath207 so that @xmath208 repeating the above steps , we find that for @xmath209 , @xmath210 and @xmath211 , @xmath210 , @xmath212 so that @xmath213 for particular values of @xmath214 , further properties of the ratio of hypergeometric function can be discussed . one particular case and few ratios of hypergeometric functions are given in section [ sec : a class of pick functions and schur functions ] with some properties . before proving such specific case , we consider another type of perturbation in @xmath0-fraction in the next section . as mentioned in section [ sec : introduction ] , the case of a single parameter @xmath76 being replaced by @xmath215 can be studied using the schur parameters . it is obvious that this is equivalent to studying the perturbed sequence @xmath216 , where @xmath217 hence , we start with a given schur function and study the perturbed carathodory function and its corresponding @xmath0-fraction . the following theorem gives the structural relation between the schur function and the perturbed one . the proof follows the transfer matrix approach , which has also been used earlier in literature ( see for example @xcite ) . [ thm : structural relations for perturbations ] let @xmath218 and @xmath219 be the @xmath53 partial numerators and denominators of the schur fraction associated with the sequence @xmath220 . if @xmath221 and @xmath222 are the @xmath53 partial numerators and denominators of the schur fraction associated with the sequence @xmath223 as defined in , then the following structural relations hold for @xmath224 , @xmath225 . @xmath226 where the entries of the transfer matrix @xmath227 are given by @xmath228 with @xmath229 let @xmath230 then the matrix relation can be written as @xmath231 with @xmath232 from , it is clear that @xmath233 defining the associated polynomials of order @xmath234 as @xmath235 we have @xmath236^{-1},\end{aligned}\ ] ] where @xmath237^{-1}$ ] denotes the matrix inverse of @xmath238 . now , using and in , we get @xmath239^{-1 } \cdot t_k(\beta_k)\cdot\omega_{k-1}(z;\alpha).\end{aligned}\ ] ] again from , @xmath240 which means @xmath241^{-1}\omega_0(z;\alpha).\end{aligned}\ ] ] using in , we get @xmath242^{-1 } \omega_0(z;\alpha)\cdot [ \omega_0(z;\alpha)]^{-1}\cdot t_k(\beta_k)\cdot\omega_{k-1}(z,\alpha),\end{aligned}\ ] ] which implies @xmath243^{t}=[t_k(\beta_k)\omega_{k-1}(z,\alpha)]^{t } \cdot [ \omega_{k}(z,\alpha)]^{-t}\cdot [ \omega_p(z,\alpha)]^{t}. \end{aligned}\ ] ] where @xmath244^{t}$ ] denotes the matrix transpose of @xmath245 . after a brief calculation , and using the relations , it can be proved that the product @xmath246^{t}\cdot \omega_{k}^{-t}(z,\alpha)$ ] precisely gives the transfer matrix @xmath227 leading to . as an important consequence of theorem , we have , @xmath247 which implies , @xmath248 this gives the perturbed schur function as , @xmath249 we next consider a non - constant schur function of the form @xmath250 where @xmath251 and have a perturbation @xmath252 . then @xmath253 the matrix entries are @xmath254z+ \alpha_0(\bar{\beta}_1-\bar{\alpha}_1),\\ % \tau_{(1,2)}&=(\beta_1-\alpha_1)z^2+[(1-\alpha_1\bar{\beta}_1)\alpha_0-(1-\bar{\alpha}_1\beta_1)\alpha_0]z+ ( \bar{\alpha}_1-\bar{\beta}_1)\alpha_0^{2},\\ % \tau_{(2,1)}&=(\alpha_1-\beta_1)(\bar{\alpha}_0)^2z^2+[(1-\beta_1\bar{\alpha}_1)\bar{\alpha}_0-(1-\alpha_1\bar{\beta}_1)\bar{\alpha}_0]z+ ( \bar{\beta}_1-\bar{\alpha}_1),\\ % \tau_{(2,2)}&=(\beta_1-\alpha_1)\bar{\alpha}_0z^2+[(1-\alpha_1\bar{\beta}_1)-(1-\bar{\alpha}_1\beta_1)\|\alpha_0\|^2]z+ ( \bar{\alpha}_1-\bar{\beta}_1)\alpha_0 . % \end{aligned}\ ] ] the transformed schur function is a rational function given by @xmath255 where @xmath256 and @xmath257 this leads to the following easy consequence of theorem [ thm : structural relations for perturbations ] . let @xmath258 , where @xmath259 denote the class of schur functions . then with the perturbation @xmath252 , the resulting schur function is the rational function given by @xmath260 we now consider an example illustrating the above discussion . [ exm : perturbed schur fractions ] consider the sequence of schur parameters @xmath35 given by @xmath261 and @xmath262 , @xmath263 . then , as in ( * ? ? ? * example 6.3 ) , the schur function is @xmath264 with @xmath265 we study the perturbation @xmath266 . for the transfer matrix @xmath227 , the following polynomials are required . @xmath267 the entries of @xmath227 are @xmath268 hence , the transformed schur function using is @xmath269 observe that @xmath28 and @xmath270 are analytic in @xmath30 with @xmath271 and @xmath272 where @xmath273 is analytic in @xmath30 with @xmath274 . further , by schwarz lemma @xmath275 for @xmath276 unless @xmath273 is a pure rotation . in such a case the range of @xmath270 is contained in the range of @xmath28 . the function @xmath270 is said to be subordinate to @xmath28 and written as @xmath277 for @xmath278 ( * ? ? ? * chapter 6 ) . we plot the ranges of both the schur functions below . in figure [ fig : subordinate - schur - functions ] , the outermost circle is the unit circle while the middle one is the image of @xmath279 under @xmath28 which is again a circle with center at 1/2 . the innermost figure is the image of @xmath279 under @xmath270 . let the carathodory function associated with the perturbed schur function @xmath280 be denoted by @xmath281 . then , using , we can write @xmath282 further , using the relation , we have @xmath283 where @xmath284 as an illustration , for the schur function @xmath264 , it is easy to verify that @xmath285 we plot these carathodory functions below . 0.4 0.4 in figures [ fig : original carat ] and [ fig : perturbed carat ] , the ranges of both the original and perturbed carathodory functions are plotted for @xmath279 . interestingly , the range of @xmath19 is unbounded ( figure [ fig : original carat ] ) which is clear as @xmath286 is a pole of @xmath19 . however @xmath287 has simple poles at 5/2 and @xmath288 and hence with the use of perturbation we are able to make the range bounded ( figure [ fig : perturbed carat ] ) . as shown in @xcite , the sequence @xmath289 satisfying the recurrence relation @xmath290 where @xmath291 and @xmath292 are the schur parameters plays an important role in the @xmath0-fraction expansion for a special class of carathodory functions . let @xmath293 correspond to the perturbed carathodory function @xmath281 . since only @xmath45 is perturbed , it is clear that @xmath294 remains unchanged for @xmath295 . the first change , @xmath296 to @xmath297 , occurs when @xmath75 is replaced by @xmath298 . consequently , @xmath299 , @xmath300 , change to @xmath301 , @xmath300 , respectively . we now show that @xmath302 can be expressed as a bilinear transformation of @xmath303 for @xmath304 . let @xmath289 be the sequence corresponding to @xmath305 and @xmath306 that to @xmath223 . then , @xmath307 where a. @xmath308 , ( j=1 ) . b. for @xmath300 , @xmath309 consider first the expression @xmath310 substituting @xmath311 and the given values of @xmath312 and @xmath313 , it simplifies to @xmath314 since @xmath315 , is proved for @xmath316 . next , let @xmath317 substituting first the given values of @xmath318 and @xmath319 , the numerator becomes @xmath320,\end{aligned}\ ] ] and then using , @xmath321 , @xmath322 where @xmath323 , thus proving for @xmath324 . the remaining part of the proof is follows by a simple induction on @xmath325 . with the condition @xmath326 , it is clear that gives analytic self - maps of the unit disk . similar to the changes in mapping properties obtained as a consequence of perturbation , it is expected that may lead to interesting results in fractal geometry and complex dynamics . since @xmath302 and @xmath303 are related by a bilinear transformation , it is clear that the expressions for @xmath327 and @xmath328 , @xmath48 , are not unique . it is known that if @xmath19 is real for real @xmath329 , then the @xmath330 are real and @xmath331 , @xmath332 . further , it is clear from that @xmath333 whenever @xmath334 . in this case , the following @xmath0-fraction is obtained for @xmath335 . @xmath336 where @xmath337 , @xmath338 , @xmath339 and @xmath340 . let the hausdorff sequence @xmath341 with @xmath342 be given so that there exists a bounded non - decreasing measure @xmath343 on [ 0,1 ] satisfying @xmath344 by ( * ? ? ? * theorem 69.2 ) , the existence of @xmath345 is equivalent to the power series @xmath346 having a continued fraction expansion of the form @xmath347 where @xmath348 and @xmath349 , @xmath154 . such functions @xmath350 are analytic in the slit domain @xmath7 and belong to the class of pick functions . we note that the pick functions are analytic in the upper half plane and have a positive imaginary part @xcite . in the next result , we characterize some members of the class of pick functions using the gap @xmath0-fraction @xmath351 . the proof is similar to that of ( * ? ? ? * theorem 1.5 ) and follows from ( * ? ? ? * lemma 3.1 ) , given earlier as ( * ? ? ? * corollary 2.1 ) . [ thm : characterising members of pick functions ] if @xmath352 with @xmath353 and @xmath354 , then the functions @xmath355 are analytic in @xmath7 and each function map both the open unit disk @xmath30 and the half plane @xmath356 univalently onto domains that are convex in the direction of the imaginary axis . we would like to note here that by a domain convex in the direction of imaginary axis , we mean that every line parallel to the imaginary axis has either connected or empty intersection with the corresponding domain @xcite , ( see also @xcite . ) . [ proof of theorem [ thm : characterising members of pick functions ] ] with the given restrictions on @xmath141 , @xmath142 and @xmath143 , @xmath357 has a @xmath0-fraction expansion and hence by ( * ? ? ? * theorem 69.2 ) , there exists a non - decreasing function @xmath358\mapsto [ 0,1]$ ] with a total increase of 1 and @xmath359 which implies @xmath360 now , if we define @xmath361 where @xmath362 , it can be easily seen that @xmath363\mapsto [ 0,1]$ ] is again a non - decreasing map with @xmath364 . further , using the contiguous relation @xmath365 we obtain @xmath366 and hence @xmath367 further , noting that the coefficient of @xmath368 in @xmath369 is @xmath370/[(c+1)(c+2)]=k_3+(1-k_3)k_2 $ ] , we define @xmath371 and find that @xmath372 \int_{0}^{1}\dfrac{\omega}{1-\sigma \omega}d\nu_2(\sigma).\end{aligned}\ ] ] finally from gauss continued fraction , we conclude that @xmath373 has a @xmath0-fraction expansion and so there exists a map @xmath374\mapsto[0,1]$ ] which is non - decreasing , @xmath375 and @xmath376 defining for @xmath377 @xmath378 so that @xmath379 , and using the fact that the coefficient of @xmath368 in @xmath373 is @xmath380 , we obtain @xmath381 \int_{0}^{1}\dfrac{\omega}{1-\sigma \omega}d\nu_4(\sigma).\end{aligned}\ ] ] thus , with @xmath382 , @xmath383 , satisfying the conditions of ( * ? ? ? * lemma 3.1 ) , ( * ? ? ? * corollary 2.1 ) , the proof of the theorem is completed . ratios of gaussian hypergeometric functions having mapping properties described in theorem @xmath384 are also found in ( * ? ? ? * theorem 1.5 ) but for the ranges @xmath385 and @xmath386 . hence for the common range @xmath353 and @xmath386 , two different ratios of hypergeometric functions belonging to the class of pick functions can be obtained leading to the expectation of finding more such ratios for every possible range . it may be noted that the ratio of gaussian hypergeometric functions in denoted here as @xmath8 has the mapping properties given in theorem @xmath384 , which is proved in ( * ? ? ? * theorem 1.5 ) . we now consider its @xmath0-fraction expansion with the parameter @xmath387 missing . using the contiguous relation and the notations used in theorems @xmath130 and @xmath388 , it is clear that @xmath389 and @xmath390 then @xmath391 then , from theorem @xmath130 , @xmath392h(2;\omega)-[1-(1-k_0)k_1z]^2}\\ & = \dfrac{c}{c - bz}-\dfrac{bczh(2;z)}{c(c - bz)h(2;z)-(c - bz)^2}\end{aligned}\ ] ] which implies @xmath393 that is @xmath394 is given as a rational transformation of a new ratio of hypergeometric functions . it may also be noted that for @xmath385 and @xmath386 , ( * ? ? ? * theorem 1.5 ) , both @xmath395 and @xmath394 will map both the unit disk @xmath30 and the half plane @xmath396 univalently onto domains that are convex in the direction of the imaginary axis . from theorem @xmath388 we obtain @xmath401 where the last equality follows from the contiguous relation hence using ( * ? ? ? 3.3 and 5.1 ) we get @xmath402 where @xmath403 is the schur function and @xmath368 and @xmath329 are related as @xmath340 . similarly , interchanging @xmath141 and @xmath142 in we obtain @xmath404 where @xmath340 . moreover , using the relation @xmath405 , @xmath48 , the related sequence of schur parameters is given by @xmath406 we note the following particular case . for @xmath407 and @xmath408 , the resulting schur parameters are @xmath409 , @xmath210 . such parameters have been considered in @xcite ( when @xmath410 ) in the context of orthogonal polynomials on the unit circle . these polynomials are known in modern literature as szeg polynomials and we suggest the interested readers to refer @xcite for further information . finally , as an illustration we note that while the schur function associated with the parameters @xmath411 is @xmath412 , that associated with the parameters @xmath413 is given by @xmath414 where @xmath340 . we remark that in this section , specific illustration of the results given in section [ sec : gap g fractions and structural relations ] are discussed leading to characterizing a class of ratio of hypergeometric functions . similar characterization of functions involving the function @xmath340 , given in section [ sec : perturbed schur parameters ] may provide some important consequences of such perturbation of @xmath0-fractions . 99 g. e. andrews , r. askey and r. roy , _ special functions _ , encyclopedia of mathematics and its applications , 71 , cambridge univ . press , cambridge , 1999 . baricz and a. swaminathan , mapping properties of basic hypergeometric functions , j. class . anal . * 5 * ( 2014 ) , no . 2 , 115128 . k. castillo , on perturbed szeg recurrences , j. math . * 411 * ( 2014 ) , no . 2 , 742752 . k. castillo , f. marcelln and j. rivero , on co - polynomials on the real line , j. math . * 427 * ( 2015 ) , no . 1 , 469483 . w. f. donoghue , jr . , the interpolation of pick functions , rocky mountain j. math . * 4 * ( 1974 ) , 169173 p. l. duren , univalent functions , grundlehren der mathematischen wissenschaften , 259 , springer - verlag , new york(1983 ) . l. garza and f. marcelln , szeg transformations and rational spectral transformations for associated polynomials , j. comput . appl . math . * 233 * ( 2009 ) , no . 3 , 730738 . m. e. h. ismail , e. merkes and d. styer , a generalization of starlike functions , complex variables theory appl . * 14 * ( 1990 ) , no . 1 - 4 , 77 - 84 . w. b. jones , o. njstad and w. j. thron , schur fractions , perron carathodory fractions and szeg polynomials , a survey , in _ analytic theory of continued fractions , ii ( pitlochry / aviemore , 1985 ) _ , 127158 , lecture notes in math . , 1199 , springer , berlin . w. b. jones , o. njstad and w. j. thron , moment theory , orthogonal polynomials , quadrature , and continued fractions associated with the unit circle , bull . london math . * 21 * ( 1989 ) , no . 2 , 113152 . w. b. jones and w. j. thron , _ continued fractions _ , encyclopedia of mathematics and its applications , 11 , addison - wesley publishing co. , reading , ma , 1980 . r. kstner , mapping properties of hypergeometric functions and convolutions of starlike or convex functions of order @xmath415 , comput . methods funct . theory * 2 * ( 2002 ) , no . 2 , 597610 . r. kstner , on the order of starlikeness of the shifted gauss hypergeometric function , j. math . * 334 * ( 2007 ) , no . 2 , 13631385 . l. lorentzen and h. waadeland , _ continued fractions with applications _ , studies in computational mathematics , 3 , north - holland , amsterdam , 1992 . e. p. merkes , on typically - real functions in a cut plane , proc . * 10 * ( 1959 ) , 863868 . o. njstad , convergence of the schur algorithm , proc . * 110 * ( 1990 ) , no . 4 , 10031007 . j. schur , ber potenzreihen dei i m inneren des einheitskreises beschrnkt sind , j. reine angewandte math . * 147 * ( 1917 ) , 205232 , * 148 * ( 1918/19 ) , 122145 . b. simon , _ orthogonal polynomials on the unit circle . part 1 _ , american mathematical society colloquium publications , 54 , part 1 , amer . soc . , providence , ri , 2005 . a. sri ranga , szeg polynomials from hypergeometric functions , proc . * 138 * ( 2010 ) , no . 12 , 42594270 . g. szeg , _ orthogonal polynomials _ , fourth edition , amer . soc . , providence , ri , 1975 . a. v. tsygvintsev , on the convergence of continued fractions at runckel s points , ramanujan j. * 15 * ( 2008 ) , no . 3 , 407413 . a. tsygvintsev , bounded analytic maps , wall fractions and @xmath416-flow , j. approx . theory * 174 * ( 2013 ) , 206219 . h. s. wall , continued fractions and bounded analytic functions , bull . * 50 * ( 1944 ) , 110119 . h. s. wall , _ analytic theory of continued fractions _ , d. van nostrand company , , new york , ny , 1948 . a. zhedanov , rational spectral transformations and orthogonal polynomials , j. comput . appl . math . * 85 * ( 1997 ) , no . 1 , 6786 .	the purpose of the present paper is to investigate some structural and qualitative aspects of two different perturbations of the parameters of @xmath0-fractions . in this context the concept of _ gap _ @xmath0-fractions is introduced . while tail sequences of a continued fraction play a significant role in the first perturbation , schur fractions are used in the second perturbation of the @xmath0-parameters that are considered . illustrations are provided using gaussian hypergeometric functions . using a particular gap @xmath0-fraction , some members of the class of pick functions are also identified .
the estimation of density derivatives has full potential for applications . this has been noted even in the first seminal papers on density estimation , as @xcite , which was also concerned with the estimation of the mode of a unimodal distribution , the value that makes zero the first density derivative . in the multivariate case , the pioneering work of @xcite showed how the estimation of the gradient vector can also be used for clustering and data filtering , leading to a substantial amount of literature on the subject , under the name of the _ mean shift algorithm_. looking further afield , @xcite made use of the mean shift idea for image analysis , and the highly - cited paper by @xcite showed how these techniques can be useful for low - level vision problems , discontinuity preserving smoothing and image segmentation . a further very popular use of the mean shift algorithm is for real - time object tracking , as described in @xcite . from the perspective of statistical data analysis , in the multidimensional context the estimation of the first and second derivatives of the density is crucial to identify significant features of the distribution , such as local extrema , valleys , ridges or saddle points . in this sense , @xcite developed methods for determining and visualizing such features in dimension two , extending previous work on scale space ideas introduced in @xcite for the univariate case ( the sizer approach ) , and the same authors also explored the application of this methodology to digital image analysis in @xcite . @xcite generalized these results for multivariate data in arbitrary dimensions and provided a novel visualization for three - dimensional data . these techniques have been widely applied recently in the field of flow cytometry ; see @xcite , @xcite or @xcite . another relatively new problem that is closely related to gradient estimation is that of finding filaments in point clouds , which has applications in medical imaging , remote sensing , seismology and cosmology . this problem is rigorously stated and analyzed in @xcite . filaments are one - dimensional curves embedded in a point process , and it can be shown that steepest ascent paths ( i.e. , the paths from each point following the gradient direction ) concentrate around them , so gradient estimation appears as a useful tool for filament detection . in this paper we focus on kernel estimators of multivariate density derivatives of arbitrary order , formally defined in section [ sec:2 ] below . as for any kernel estimator , it is well known that the crucial factor that determines the performance of the estimator in practice is the choice of the bandwidth matrix . in the multivariate setting there are several levels of sophistication at the time of specifying the bandwidth matrix to be used in the kernel estimator ( see * ? ? ? * chapter 4 ) . the most general bandwidth type consists of a symmetric positive definite matrix ; it allows the kernel estimator to smooth in any direction whether coordinate or not . this general class of bandwidth matrices can be constrained to consider positive definite diagonal matrices , allowing for different degrees of smoothing along each of the coordinate axis , or even further to consider a bandwidth matrix involving only a positive scalar multiple of the identity matrix , meaning that the same smoothing is applied to every coordinate direction . as noted in @xcite in the density estimation context , the single - parameter class should not be used for unscaled data or , as stated by @xcite in terms of feature space analysis , to use this bandwidth class at least the validity of an euclidean metric for the feature space should be previously checked . the simpler parameterizations are in general more widely used than the unconstrained counterpart for two reasons : first , in practice they need less smoothing parameters to be tuned , and second , due to the difficulties encountered in the mathematical analysis of estimators with unconstrained bandwidths . however , @xcite provided a detailed error analysis of kernel density derivative estimators with unconstrained bandwidths and showed that the use of the simpler parameterizations can lead to a substantial loss in terms of efficiency , and that this problem becomes more and more important as the order of the derivative to be estimated increases . @xcite also proposed an optimal bandwidth selector for the normal case , but they did not develop more sophisticated data - driven choices of the bandwidth matrix with applicability to more general densities , which is crucial to make density derivative estimation useful in practice . along the same lines , @xcite argue that most existing bandwidth selection methods for the mean shift algorithm , all of them for the single - parameter class of bandwidths , are based on empirical arguments . in the univariate case there exist some approaches to bandwidth selection for density derivative estimation : @xcite introduced a cross validation method and showed its optimality ; @xcite derived the relative rate of convergence of this method and also for a plug - in proposal ; @xcite studied two root @xmath0 selectors in the fourier domain , and more recently @xcite focused on the smoothed cross validation bandwidth selector for the density derivative . in the multivariate case , however , the issue of automatic bandwidth selection for density derivative estimation has remained largely unexplored . given the smaller body of multivariate density estimation research as compared to their univariate cousins , it is not surprising that multivariate density derivative estimation suffers equally ( if not more so ) from a lack of solid results . to the best of our knowledge , in the literature the only published approaches to bandwidth selection for multivariate kernel estimation of density derivatives are the recent papers @xcite and @xcite , but both focus exclusively on the first derivative . this paper proposes three new methods for unconstrained bandwidth matrix selection for the multivariate kernel density derivative estimator , and explores their applicability to other related statistical problems . in section [ sec:2 ] , we introduce the mathematical framework for the analysis of multivariate derivatives . in section [ sec:3 ] we show that the relative rate of convergence of these unconstrained selectors is the same as for the classes of simpler bandwidth matrices , so that from an asymptotic point of view our methods can be as successful as ( and more flexible than ) those needing less smoothing parameters . the finite - sample behaviour of the new bandwidths is investigated in section [ sec:5 ] , and their application to develop new data - driven nonparametric clustering methods via the mean shift algorithm is explored in section [ sec : ms ] , and for feature significance in section [ sec : feature ] . finally , the proofs of the results are given in an appendix . the problem of estimating the @xmath1-th derivative of a multivariate density is considered in this section . from a multivariate point of view , the @xmath1-th derivative of a function is understood as the set of all its partial derivatives of order @xmath1 , rather than just one of them . notice that , for instance , in a multivariate taylor expansion of order @xmath1 all of the partial derivatives of order @xmath1 are needed to compute the @xmath1-th order term . or , in another related example , all the second order partial derivatives are involved in the computation of the hessian matrix . all the @xmath1-th partial derivatives can be neatly organized into a single vector as follows : if @xmath2 is a real @xmath3-variate density function and @xmath4 , denote by @xmath5 the first derivative ( gradient ) operator . all the second order partial derivatives can be organized into the hessian matrix @xmath6 , and the hessian operator can be formally written as @xmath7 if the usual convention @xmath8 is taken into account . for @xmath9 , however , it is not that clear how to organize the set containing all the @xmath10 partial derivatives of order @xmath1 . here we adopt the unified approach used in @xcite or ( * ? ? ? * section 1.4 ) , where the @xmath1-th derivative of @xmath2 is defined to be the vector @xmath11 . in the previous equation @xmath12 denotes the @xmath1-th kronecker power of the operator @xmath13 ; see , e.g. , @xcite for the definition of the kronecker product . naturally , @xmath14 , @xmath15 and , for example , @xmath16 , where @xmath17 denotes the operator which concatenates the columns of a matrix into a single vector . here we study the problem of estimating the @xmath1-th derivative @xmath18 from a sample @xmath19 of independent and identically distributed random variables with common density @xmath2 . the usual kernel estimator of @xmath2 is defined as @xmath20 , where the kernel @xmath21 is a spherically symmetric density function , the bandwidth @xmath22 is a symmetric positive definite matrix and @xmath23 . thus , the most straightforward estimator of @xmath18 is just the @xmath1-th derivative of @xmath24 , given by @xmath25 , where the roles of @xmath21 and @xmath22 can be separated for implementation purposes by noting that @xmath26 , as shown in @xcite , where for any matrix @xmath27 it is understood that @xmath28 and @xmath29 . see , however , @xcite for other possible estimators in the univariate context . for the density estimation case ( @xmath30 ) , @xcite showed that the use of bandwidths belonging to the class @xmath31 , with @xmath32 the @xmath33 identity matrix , or the class @xmath34 , may lead to dramatically less efficient estimators than those based on bandwidth matrices drawn from @xmath35 , the space of all positive definite symmetric matrices . moreover @xcite showed that the issue of efficiency loss is even more severe for @xmath36 . so the development of unconstrained bandwidth selectors for density derivative estimation , which is achieved in this paper , may also represent an important improvement in practice . to measure the error committed by the kernel estimator for the sample at hand it is natural to consider the integrated squared error ( ise ) , defined as @xmath37 where @xmath38 denotes the usual euclidean norm . this quantity depends on the data , so it is common to consider the mean integrated squared error @xmath39 $ ] as a non - stochastic measure of error , and its minimizer @xmath40 as the non - stochastic optimal bandwidth choice . a more detailed discussion of the advantages and disadvantages of the ise and mise viewpoints can be found in @xcite . standard calculations lead to the integrated variance plus integrated squared bias decomposition @xmath41 , where @xmath42d{{\boldsymbol x}}$ ] and @xmath43-{\mathsf{d}}^{\otimes r}f({{\boldsymbol x}})\\|^2d{{\boldsymbol x}}$ ] . by expanding each of these two terms , @xcite showed that a more explicit form of the mise is given by @xmath44 where @xmath45 , and @xmath46 for a vector valued function @xmath47 and a real valued function @xmath48 , with @xmath49 standing for a component - wise application of the convolution operator . moreover , writing @xmath50 , under some smoothness assumptions @xcite also showed that an asymptotic approximation of the mise is given by @xmath51 and that the minimizer of this amise function , @xmath52 , has entries of order @xmath53 , leading to a minimum achievable amise of order @xmath54 . although these expressions provide an insightful error analysis of multivariate kernel density derivative estimators , they are not directly implementable as software since they all involve the unknown density @xmath2 . in the next section , we examine strategies to estimate these unknown quantities which lead to optimal data - based selectors for density derivative estimation . in this section we propose three new methods to select the bandwidth matrix for kernel density derivative estimation from the data and study their asymptotic properties . these methods are inspired by the cross validation , plug - in and smoothed cross validation methodologies for the estimation of the density in the univariate case , hence their names . cross validation ( cv ) techniques for bandwidth selection for univariate density estimation were introduced in @xcite and @xcite , and studied in detail in seminal papers like @xcite , @xcite and @xcite . they can be motivated in terms of either ise - oriented or mise - oriented considerations . in the case of multivariate density derivative estimation notice that , for a random variable @xmath55 having density @xmath2 and independent of @xmath19 , using integration by parts it is possible to write @xmath56\bigg\}\\ & \quad+\operatorname{tr}{\mathbf{r}}({\mathsf{d}}^{\otimes r}f)\end{aligned}\ ] ] provided that the kernel @xmath21 is sufficiently smooth . the last term is irrelevant for minimizing concerns , and the two first terms can be unbiasedly estimated by @xmath57^{-1}\sum_{i\neq j}{\mathsf{d}}^{\otimes 2r}k_{{\bf h}}({{\bf x}}_i-{{\bf x}}_j)\bigg\}\end{aligned}\ ] ] where @xmath58 denotes the kernel estimator based on the sample with the @xmath59-th observation deleted . from the mise point of view notice that , for any smooth enough function @xmath60 , @xmath61,\end{aligned}\ ] ] so that @xmath62 can be unbiasedly estimated by @xmath63^{-1}{\operatorname{vec}}^\top{{\bf i}}_{d^r}\sum_{i\neq j}{\mathsf{d}}^{\otimes 2r}l_{{\bf h}}({{\bf x}}_i-{{\bf x}}_j).\ ] ] therefore , in view of ( [ exactmise ] ) , @xmath64 can be unbiasedly estimated by @xmath65^{-1}{\operatorname{vec}}^\top{{\bf i}}_{d^r}\sum_{i\neq j}^n\big\{(1-n^{-1}){\mathsf{d}}^{\otimes 2r}\bar{k}_{{\bf h}}-2{\mathsf{d}}^{\otimes 2r}k_{{\bf h}}\big\}({{\bf x}}_i-{{\bf x}}_j),\end{aligned}\ ] ] where @xmath66 . to check that these two cv expressions coincide , take into account that @xmath67 , so that using some properties of the kronecker product and the vec operator ( see , e.g. , * ? ? ? * ) , the sum of the diagonal terms in the first @xmath68 expression equals @xmath69 where the third line follows by noting that @xmath70 . surely the simplest formulation for cv ( useful for implementation purposes ) is @xmath71^{-1}\sum_{i\neq j}{\mathsf{d}}^{\otimes 2r}k\big({{\bf h}}^{-1/2}({{\bf x}}_i-{{\bf x}}_j)\big)\bigg\}.\end{aligned}\ ] ] we denote by @xmath72 the bandwidth matrix in @xmath73 which minimizes @xmath74 . plug - in ( pi ) bandwidth selection techniques are based on estimating the unknown quantities that appear in an asymptotic error formula and minimizing the resulting empirical criterion . basic plug - in selectors for univariate density estimation are described in @xcite and @xcite . in the multivariate case , introducing the vector integrated density derivative functional , defined as @xmath75 allows us to rewrite the amise formula ( [ amise ] ) for the @xmath1-th derivative as @xmath76 thus , the plug - in bandwidth selector @xmath77 is defined to be the bandwidth in @xmath78 minimizing the criterion @xmath79 where @xmath80 is a suitable estimator of @xmath81 . @xcite analyzed the problem of estimating @xmath82 for an arbitrary @xmath1 . noting that @xmath83 $ ] , they proposed the kernel estimator @xmath84 using a kernel @xmath60 with pilot bandwidth @xmath85 , possibly different from @xmath21 and @xmath22 . for the selection of the pilot bandwidth matrix @xmath85 , the same authors showed that the leading term of the mean squared error @xmath86 $ ] is given by the squared norm of the asymptotic bias vector @xmath87 so that the asymptotically optimal choice of the pilot bandwidth for the estimation of @xmath82 is @xmath88 , which depends on @xmath89 . hence , to select the pilot bandwidth @xmath85 from the data we could substitute @xmath89 by another kernel estimator in ( [ eq : omega_pi ] ) and minimize the squared norm of the resulting vector , but of course then the optimal bandwidth for the kernel estimator of @xmath90 depends on @xmath81 , and so on . the usual strategy to overcome this problem is to use an @xmath91-stage algorithm as described in @xcite , involving @xmath91 successive kernel functional estimations with the initial bandwidth matrix chosen on the basis of a normal scale approximation . the resulting bandwidth obtained by minimizing @xmath92 when @xmath81 is estimated using an @xmath91-stage algorithm is called an @xmath91-stage plug - in bandwidth selector for the @xmath1-th derivative . the smoothed cross validation ( scv ) methodology for univariate density estimation was introduced in @xcite , and a thorough study of this technique was made in @xcite . however , it has not been until recently that its multivariate counterpart has been developed , in @xcite and @xcite , and its use for univariate density derivative estimation has been explored ( see * ? ? ? * ) . a possible derivation of the scv criterion for the problem of multivariate density derivative estimation is based on the approximation of the mise obtained by replacing the exact integrated variance in equation ( [ exactmise ] ) by its asymptotic approximation ( the first term ) , while keeping the exact form for the integrated squared bias , so that @xmath93 with @xmath94 where @xmath95 ( here @xmath96 denotes the dirac delta function ) and @xmath97 . the scv criterion is obtained by replacing the unknown target @xmath98 in the mise2 formula with a pilot estimator @xmath99 , leading to @xmath100 where @xmath101 . when all the @xmath102 are distinct and the diagonal terms ( @xmath103 ) are omitted in the previous sum it can be shown , using the properties of the dirac delta function ( see , e.g. , * ? ? ? * chapter i.2 ) , that the scv criterion coincides with the cv criterion for @xmath104 . the minimizer of @xmath105 is defined to be @xmath106 . its value depends on the pilot selector @xmath85 . @xcite showed that in the case @xmath30 the leading term of the mean squared error @xmath107 is given by the squared norm @xmath108 where @xmath109 is the same as the aforementioned @xmath110 except that @xmath60 is replaced by @xmath111 . thus it is straightforward to define , analogously to the plug - in algorithm , the required optimal @xmath112-th stage pilot bandwidth of an @xmath91-stage scv selector . let @xmath113 be an arbitrary data - based bandwidth selector , built up on the basis of an estimated criterion @xmath114 . following @xcite , @xmath115 is said to converge to @xmath116 at relative rate @xmath117 if @xmath118 where @xmath119 denotes element - wise order in probability and @xmath120 is the @xmath121 matrix of ones . this order in probability statement can be difficult to derive directly . the next lemma provides a more tractable indirect method of calculating convergence rates . [ lem : asymhr ] suppose that assumptions ( a1)(a3 ) given in the appendix hold . the discrepancy @xmath122 is asymptotically equivalent to @xmath123({{\bf h}}_{\mise , r})$ ] , where @xmath124 is shorthand for @xmath125 . furthermore , the relative rate of convergence of @xmath126 is @xmath117 if @xmath127({{\bf h}}_{\mise , r } ) \rbrace \operatorname{\boldsymbol{\mathbb{e}}}\lbrace { \mathsf{d}}_{{\bf h}}[\widehat{\mathrm{mise}}_r-\mathrm{mise}_r]({{\bf h}}_{\mise , r } ) \rbrace^\top \\ & + \operatorname{var}\lbrace { \mathsf{d}}_{{\bf h}}[\widehat{\mathrm{mise}}_r-\mathrm{mise}_r]({{\bf h}}_{\mise , r } ) \rbrace = o(n^{-2\alpha } { \mathbf{j}}_{d^2 } ) { \operatorname{vec}}{{\bf h}}_{\mise , r } { \operatorname{vec}^\top}{{\bf h}}_{\mise , r}.\end{aligned}\ ] ] the convergence rates of the three bandwidth selectors considered here are given in the following theorem , whose proof is deferred to the appendix . [ reroc ] suppose that assumptions ( a1)(a5 ) given in the appendix hold . the relative rate of convergence to @xmath116 is @xmath128 for the cross validation selector @xmath129 , and @xmath130 for the plug - in selector @xmath131 and the smoothed cross validation selector @xmath106 when @xmath132 . @xcite computed the relative rate of convergence for the cv and pi selectors for the estimation of a single partial derivative , using a single - parameter bandwidth matrix ( i.e. , @xmath133 ) . the previous theorem shows that the unconstrained cv bandwidth attains the same rate as its constrained counterpart , yet with added flexibility that should be captured in the constant coefficient of the asymptotic expression , although the computation of an explicit form for this coefficient does not seem possible in general . the convergence rate of the pi selector is @xmath134 within the single - parameter bandwidth class @xmath135 , yielding a slightly faster convergence to the optimal constrained bandwidth . as explained in @xcite for the density case , this is due to the fact that the very special cancellation in the bias term which is achievable when using a single - parameter bandwidth is not possible in general for the unconstrained estimator . nevertheless , the aforementioned papers showed that this slight loss in convergence rate terms is negligible in practice as compared with the fact that the targeted constrained optimal bandwidth is usually much less efficient than the unconstrained one ( see also section [ sec:5 ] below ) . theorem [ reroc ] also shows that the similarities noted in @xcite about the asymptotic properties of the pi and scv methods for the density estimation problem persist for @xmath136 , since both selectors exhibit the same relative rate of convergence . @xcite exemplified how slow is the rate @xmath137 of the cv selector for @xmath138 , @xmath30 by noting that @xmath0 has to be as large as @xmath139 so that @xmath140 . in the same spirit , to compare the rates obtained in theorem [ reroc ] , table [ rates - ratio ] shows the values of @xmath128 ( cv ) and @xmath130 ( pi and scv ) divided by @xmath141 , that is the rate for the cv selector @xmath142 which is used as a base case , for all the different combinations of @xmath143 , @xmath144 and @xmath145 . ratios which are lower than 1 indicate the rate is faster than the base case , and ratios greater than 1 a slower rate . for @xmath146 , these ratios in table [ rates - ratio ] tend to be greater than 1 , indicating that using this sample size will lead to a deteriorating convergence rate . on the other hand for the larger sample sizes , @xmath147 , these ratios tend to be less than 1 . this implies that convergence rates better than the cv selector for bivariate density estimation can be attained , even with higher dimensions and higher order derivatives , provided that sufficiently large ( although still realistic ) sample sizes are used . of course this comparison only takes into account the asymptotic order of the convergence rates by ignoring the associated coefficients since explicit formulas for the latter are not available for @xmath148 . the finite sample behaviour of the bivariate case for moderate sample sizes is examined more closely in the next section . .comparison of the relative rate of convergence for the cv , pi and scv selectors . for each combination of @xmath1 , @xmath0 and @xmath3 in the table , the left entry in the corresponding cell shows @xmath128 ( cv selector ) and the right entry @xmath130 ( pi and scv selectors ) divided by @xmath141 ( i.e. the rate for the cv selector with @xmath142 ) . [ cols="<,<,^,^,^,^,^,^,^,^ " , ] in view of table [ tab : ari ] , none of the methods compared is uniformly the best . in the group of the mean shift procedures , the use of the pi bandwidth seems to exhibit the best overall performance . the cv choice can be rated second best , with similar or even slightly ( but not significantly ) better average ari in some cases . the scv bandwidth shows an unexpectedly inferior performance for the normal mixture models , but it has an acceptable behaviour for the models with non - standard cluster shapes . finally , the normal scale rule nr is clearly inferior in four out of the five models , but it performs surprisingly well for the broken ring model ; since it is the least intensive method in computational terms , it could be useful at least to provide a quick initial analysis , especially in higher dimensions . the comparison with the parametric method mclust followed the expected guidelines : for the normal mixture models mclust showed good results , especially for the difficult quadrimodal density , but it seems unable to adapt itself to the non - standard cluster shape situations . on the contrary , clues is not very powerful for a standard setup with ellipsoidal clusters , but seems to performs reasonably well for non - standard problems . finally , pdfc shows remarkable results in the simulation study , in spite of the ad hoc choice of the bandwidth in which it is based , and its performance is comparable to that of the best mean shift procedure , with the only exception of the eye model . surely a more careful study of the bandwidth selection problem would improve the quality of the pdfc method further . the mean shift algorithm in conjunction with the new proposed bandwidth selection rules was also applied to some real data sets . it is well - known that the kernel density estimator tends to produce spurious bumps ( i.e. , unimportant modes caused by a single observation ) in the tails of the distribution , and that this problem seems enhanced in higher dimensions , due to the empty space phenomenon and the curse of dimensionality ( see , for instance , * ? ? ? * chapter 4 ) . for real data sets , this may result in a number of data points forming singleton clusters after applying the mean shift algorithm . furthermore , in some applications the researcher may be interested in forming more homogeneous groups so that , say , insignificant groups of size less than @xmath149 of the biggest group are not allowed in the outcome of the clustering algorithm . this goal can be achieved as follows : apply the mean shift algorithm to the whole data set and identify all the data points forming groups of size less than @xmath149 of the biggest group , then leave those singular data points out of the estimation process in the mean shift algorithm and re - compute the data - based bandwidth and the density and density gradient estimators in ( [ eq : meanshift ] ) using only non - singular data points . since the mean shift algorithm produces a partition of the whole space , these left - out data points can be naturally assigned to any of the corresponding newly obtained clusters . if this new assignment again contains insignificant clusters then iterate the process until the eventual partition satisfies the desired requirements . this correction is similar ( although a little different ) to the stage called merging clusters based on the coverage rate " in @xcite , and will be referred henceforth as _ correction for insignificant groups_. the _ e. coli _ data set is provided by the uci machine learning database repository @xcite . the original data were contributed by kenta nakai at the institute of molecular and cellular biology of osaka university . the data represent seven features calculated from the amino acid sequences of @xmath150 e.coli proteins , classified in eight classes according to their localization sites , labeled iml ( 2 observations ) , oml ( 5 ) , ims ( 2 ) , om ( 20 ) , pp ( 52 ) , imu ( 35 ) , i m ( 77 ) , cp ( 143 ) . a more detailed description of this data set can be found in @xcite . since two of the original seven features are binary variables , only the remaining five continuous variables ( @xmath151 ) , scaled to have unit variance , were retained for the cluster analysis . the number of groups identified by the mean shift procedure with correction for insignificant groups ( using @xmath152 as a default ) was 5 for pi and scv bandwidths , which is the natural choice if the insignificant clusters iml , oml and ims are merged into bigger groups . the mean shift algorithm found 6 groups using the nr bandwidth and 7 with the cv bandwidth . since in this example the true cluster membership is available from the original data , it is also possible to compare the performance of the methods using the ari . the aris for these configurations were 0.63 ( nr bandwidth ) , 0.671 ( cv ) , 0.667 ( pi ) and 0.559 ( scv ) . in contrast , clues and pdfc indicated a severely underestimated number of groups in the data , namely 3 and 2 , respectively , and whereas clues obtains a remarkably high ari anyway ( 0.697 ) , the performance of pdfc is poor for this data set in ari terms ( 0.386 ) . mclust also gives a reasonable answer , with 6 groups and an ari of 0.642 . these data were introduced in @xcite , and consist of eight chemical measurements on @xmath153 olive oil samples from three regions of italy . the three regions r1 , r2 and r3 are further divided into nine areas , with areas a1 ( 25 observations ) , a2 ( 56 ) , a3 ( 206 ) and a4 ( 36 ) in region r1 ( totalling 323 observations ) ; areas a5 ( 65 ) and a6 ( 33 ) in region r2 ( totalling 98 ) ; and areas a7 ( 50 ) , a8 ( 50 ) and a9 ( 51 ) in region r3 ( totalling 151 ) . detailed cluster analyses of this data set are given in @xcite and @xcite . taking into account the compositional nature of these data , they were transformed following the guidelines in the latter reference , first dealing with the effect of rounding zeroes when the chemical measurement was below the instrument sensitivity level and then applying the additive log - ratio transform to place the data in a 7-dimensional euclidean space ( see * ? ? ? * for a recent monograph on compositional data ) . then , cluster analysis was carried out over the first five principal components of the scaled euclidean variables . the results of the analysis indicated that whereas some methods seemed to target the partition of the data into major regions , others tried hard to discover the sub - structure of areas . this was clearly recognized when the aris of the groupings were computed either with respect to one classification or the other . naturally , if a method produced a grouping which was accurate with respect to major regions , it had lower ari with respect to the division into areas . clues , pdfc and the mean shift algorithm using the nr bandwidth clearly favoured grouping the data into major categories . the pdfc method obtained a remarkable ari of 0.841 by clustering the data into 3 groups , whereas clues only found 2 groups resulting in an ari of 0.680 . using the nr bandwidth the mean shift algorithm achieved an ari of 0.920 with respect to the true grouping into major regions ; it correctly identified all the data points in regions r1 and r2 , although region r3 appeared divided into several subregions . in contrast , mclust and the mean shift algorithm combined with all the more sophisticated bandwidth selectors tended to produce groupings closer to the assignment into smaller areas . mclust showed the existence of 8 groups and achieved an ari of 0.739 with respect to the true distribution into areas . the mean shift analyses with the cv , pi and scv bandwidths all found 7 groups , leading to aris of 0.741 ( cv bandwidth ) , 0.791 ( pi ) and 0.782 ( scv ) . it is not always easy to interpret visually estimates of multivariate derivatives . to assist us , we use the significant negative curvature regions of @xcite , defined as the set containing the values of @xmath154 such that the null hypothesis that the hessian @xmath155 is positive definite is significantly rejected . the appropriate kernel test statistic , null distribution and adjustment for multiple testing is outlined in @xcite and implemented in the ` feature ` library in ` r ` . significant negative curvature regions corresponds to a modal region in the density function , and hence a local maxima in data density . these authors focused on the scale space approach of smoothing and so did not develop optimal bandwidth selectors for their density derivative estimates . here , we compare the significant curvature regions obtained using a usual @xmath30 bandwidth selector to those with an @xmath156 optimal bandwidth in figure [ fig : earthquake ] on the earthquake data from @xcite . the recorded measurements are the latitude and longitude ( in degrees ) and depth ( in km ) of epicenters of 510 earthquakes . here , negative latitude indicates west of the international date line , and negative depth indicates distances below the earth s surface . the depth is transformed using log(depth ) . for these transformed data , we use pi selectors @xmath157 and @xmath158 and scv selectors @xmath159 and @xmath160 . as expected from asymptotic theory , bandwidths for hessian estimation are larger in magnitude than bandwidths for density estimation . moreover only the central modal region is present using @xmath157 , whereas with @xmath158 , the three local modal regions are more clearly delimited from the surrounding space , confirming the three modes obtained with subjective bandwidth selection by @xcite . . ( upper right ) plug - in selector @xmath156 . ( lower left ) scv selector @xmath30 . ( lower right ) scv selector @xmath156 . the significant curvature regions or modal regions are more clearly delimited from the surrounding scatter point cloud with the selectors corresponding to second derivative.,title="fig:",scaledwidth=50.0% ] . ( upper right ) plug - in selector @xmath156 . ( lower left ) scv selector @xmath30 . ( lower right ) scv selector @xmath156 . the significant curvature regions or modal regions are more clearly delimited from the surrounding scatter point cloud with the selectors corresponding to second derivative.,title="fig:",scaledwidth=50.0% ] + . ( upper right ) plug - in selector @xmath156 . ( lower left ) scv selector @xmath30 . ( lower right ) scv selector @xmath156 . the significant curvature regions or modal regions are more clearly delimited from the surrounding scatter point cloud with the selectors corresponding to second derivative.,title="fig:",scaledwidth=50.0% ] . ( upper right ) plug - in selector @xmath156 . ( lower left ) scv selector @xmath30 . ( lower right ) scv selector @xmath156 . the significant curvature regions or modal regions are more clearly delimited from the surrounding scatter point cloud with the selectors corresponding to second derivative.,title="fig:",scaledwidth=50.0% ] + * acknowledgments . * grant mtm2010 - 16660 ( both authors ) from the spanish ministerio de ciencia e innovacin , and various fellowships ( second author ) from the institut curie , france , and the institute of translational sciences , france have supported this work . henceforth the following assumptions are made : 1 . @xmath21 is a symmetric @xmath3-variate density such that @xmath161 and all its partial derivatives up to order @xmath162 are bounded , continuous and square integrable . 2 . @xmath2 is a density function with all its partial derivatives up to order @xmath163 bounded , continuous and square integrable . 3 . @xmath164 is a sequence of bandwidth matrices such that all entries of @xmath165 and @xmath22 tend to zero as @xmath166 . these do not form a minimal set of assumptions , but they serve as useful starting point for the results that we subsequently develop . besides , in this section integrals without any integration limits are assumed to be integrated over the appropriate euclidean space . we also assume that suitable regularity conditions are satisfied so that the exchange of term - by - term integration and differentiation of taylor expansions are well - defined . reasoning as in lemma 1 in @xcite , it follows that @xmath122 is asymptotically equivalent to @xmath167^{-1}{\mathsf{d}}_{{\bf h}}[\widehat{\mathrm{mise}}_r-\mathrm{mise}_r]({{\bf h}}_{\mise , r})$ ] , where @xmath168 denotes the hessian operator corresponding to @xmath124 . therefore , it suffices to show that @xmath169 . but for any @xmath22 with entries of order @xmath53 , as @xmath116 , the results in @xcite imply that @xmath170 is equivalent to @xmath171 and , moreover , the smoothness assumptions ensure that @xmath172 is of the same order as @xmath173 . and it is not hard to show that for the asymptotic integrated squared bias term we have @xmath174 and similarly for the asymptotic integrated variance , thus finishing the proof . lemma [ lem : asymhr ] shows that @xmath175 is asymptotically equivalent to @xmath176({{\bf h}}_{\mise , r})$ ] . since @xmath177=\mise_r({{\bf h}})-\operatorname{tr}{\mathbf{r}}({\mathsf{d}}^{\otimes r}f)$ ] for all @xmath22 , it follows that the order of @xmath175 is given by the ( root ) order of @xmath178({{\bf h}}_{\mise , r})\big\}\\ & \sim\operatorname{var}\bigg\{[n(n-1)]^{-1}\sum_{i\neq j}^n{\mathsf{d}}_{{\bf h}}\big[{\operatorname{vec}}^\top({{\bf h}}^{-1})^{\otimes r}({\mathsf{d}}^{\otimes 2r}\tilde k)_{{\bf h}}({{\bf x}}_i-{{\bf x}}_j)\big]\big\vert_{{{\bf h}}={{\bf h}}_{\mise , r}}\bigg\},\end{aligned}\ ] ] where @xmath179 . so denoting @xmath180 $ ] , by standard @xmath181-statistics theory the previous variance is of the same order as @xmath182 where @xmath183\\ { \mathbf{\xi}}_2&=\operatorname{\boldsymbol{\mathbb{e}}}[\boldsymbol{\varphi}_{{\bf h}}({{\bf x}}_1-{{\bf x}}_2)\boldsymbol{\varphi}_{{\bf h}}({{\bf x}}_1-{{\bf x}}_2)^\top]\\ { \mathbf{\xi}}_0&=\operatorname{\boldsymbol{\mathbb{e}}}[\boldsymbol{\varphi}_{{\bf h}}({{\bf x}}_1-{{\bf x}}_2)]\operatorname{\boldsymbol{\mathbb{e}}}[\boldsymbol{\varphi}_{{\bf h}}({{\bf x}}_1-{{\bf x}}_2)]^\top\end{aligned}\ ] ] with @xmath22 of the order of @xmath116 , namely having all its entries of order @xmath53 . the following lemma provides an explicit expression for the function @xmath184 that will be helpful to evaluate @xmath185 . [ lem : varphi ] the function @xmath184 can be explicitly expressed as @xmath186 where the function @xmath187 is given by @xmath188 and the matrices @xmath189 , @xmath190 are defined as @xmath191\end{aligned}\ ] ] where we understand that @xmath192 . since @xmath193 , its differential is decomposed into three terms @xmath194 from @xcite , the differentials involved in the first two terms can be expressed as @xmath195 d { \operatorname{vec}}{{\bf h}},\end{aligned}\ ] ] where @xmath196 is a matrix such that @xmath197 . for the third term , @xmath198 since @xmath199{\operatorname{vec}}{{\bf h}}^{\otimes -r}={\operatorname{vec}}\big({{\bf i}}_d[{\mathsf{d}}({\mathsf{d}}^\top)^{\otimes 2r}]{\operatorname{vec}}{{\bf h}}^{\otimes -r}\big)=({\operatorname{vec}^\top}{{\bf h}}^{\otimes - r}\otimes{{\bf i}}_d){\mathsf{d}}^{\otimes 2r+1}$ ] . finally , using @xmath200 from @xcite , it follows that @xmath201 . thus the derivative reads @xmath202 the central factors of the third term on the right hand side can be rewritten as @xmath203\\ & = [ { \operatorname{vec}^\top}{{\bf h}}^{\otimes -r } \otimes ( { { \bf h}}^{1/2 } \otimes { { \bf h}}^{1/2 } + { { \bf i}}_d \otimes { { \bf h}})^{-1 } ] ( { { \bf i}}_{d^{2r } } \otimes { { \bf h}}^{-1/2}{{\boldsymbol x}}\otimes { { \bf i}}_d),\end{aligned}\ ] ] as desired . we now return to the task of finding the asymptotic order of @xmath204 , @xmath205 and @xmath206 . for that , some preliminary notation is needed . for any real function @xmath48 we denote its vector moment of order @xmath207 as @xmath208 . for instance , @xcite showed that @xmath209 , @xmath210 and @xmath211 , where @xmath212 denotes the symmetrizer matrix of order @xmath1 ( see * ? ? ? * ) , defined as the ( only ) matrix such that pre - multiplying a kronecker product of any @xmath1 vectors in @xmath213 by @xmath214 results in the average of all possible permutations of the @xmath1-fold product . we also introduce here the notation @xmath215 for the commutation matrix of order @xmath216 @xcite . so taking this into account , for the calculation of the asymptotic order of @xmath204 , a fourth order taylor expansion of @xmath217 , in the form of ( * ? ? ? * theorem 1.4.8 ) or @xcite , gives @xmath218{\mathsf{d}}^{\otimes 2r+p}f({{\boldsymbol x}})d{{\boldsymbol z}}\\ & = ( { { \bf h}}^{1/2})^{\otimes 2r}\sum_{p=0}^4\frac{(-1)^p}{p!}\big[{{\bf i}}_{d^{2r}}\otimes({{\boldsymbol \mu}}_p(\tilde k)^\top ( { { \bf h}}^{1/2})^{\otimes p})\big]{\mathsf{d}}^{\otimes 2r+p}f({{\boldsymbol x}})\\ & = -({{\bf h}}^{1/2})^{\otimes 2r}{\mathsf{d}}^{\otimes 2r}f({{\boldsymbol x}})+\tfrac{1}{4}m_2(k)^2({{\bf h}}^{1/2})^{\otimes 2r}\big[{{\bf i}}_{d^{2r}}\otimes((\operatorname{vec}^\top { { \bf h}})^{\otimes 2}\boldsymbol{\mathcal s}_{d,4})\big]{\mathsf{d}}^{\otimes 2r+4}f({{\boldsymbol x}})\\ & = -({{\bf h}}^{1/2})^{\otimes 2r}{\mathsf{d}}^{\otimes 2r}f({{\boldsymbol x}})+\tfrac{1}{4}m_2(k)^2\big[({{\bf h}}^{1/2})^{\otimes 2r}\otimes(\operatorname{vec}^\top { { \bf h}})^{\otimes 2}\big]{\mathsf{d}}^{\otimes 2r+4}f({{\boldsymbol x}}),\end{aligned}\ ] ] therefore , since @xmath219 and @xmath220 , we obtain @xmath221\nonumber\\ & \sim{\mathsf{d}}_{{\bf h}}\big\{\tfrac{1}{4}m_2(k)^2\big[{\operatorname{vec}}^\top { { \bf i}}_{d^r}\otimes(\operatorname{vec}^\top { { \bf h}})^{\otimes 2}\big]{\mathsf{d}}^{\otimes 2r+4}f({{\boldsymbol x}})\big\}\nonumber\\ & = \tfrac{1}{4}m_2(k)^2\big({\operatorname{vec}}^\top { { \bf i}}_{d^r}\otimes{{\bf i}}_{d^2}\otimes{\operatorname{vec}}^\top { { \bf h}}\big)[{{\bf i}}_{d^{2r}}\otimes({{\bf i}}_{d^2}+{\mathbf{k}}_{d^2,d^2})]{\mathsf{d}}^{\otimes 2r+4}f({{\boldsymbol x}})\nonumber\\ & = \tfrac{1}{2}m_2(k)^2\big({\operatorname{vec}}^\top { { \bf i}}_{d^r}\otimes{{\bf i}}_{d^2}\otimes{\operatorname{vec}}^\top { { \bf h}}\big){\mathsf{d}}^{\otimes 2r+4}f({{\boldsymbol x}}).\label{phihf}\end{aligned}\ ] ] using this , @xmath222&=\int\boldsymbol\varphi_{{\bf h}}f({{\boldsymbol x}})f({{\boldsymbol x}})d{{\boldsymbol x}}\sim\tfrac{1}{2}m_2(k)^2\big({\operatorname{vec}}^\top { { \bf i}}_{d^r}\otimes{{\bf i}}_{d^2}\otimes{\operatorname{vec}}^\top { { \bf h}}\big)\boldsymbol\psi_{2r+4}\end{aligned}\ ] ] and @xmath223 similarly , @xmath224 finally , note that @xmath225 $ ] , where @xmath226 , which also depends on @xmath22 through @xmath27 and @xmath227 . besides , @xmath228=\iint(\boldsymbol\varphi\boldsymbol\varphi^\top)({{\boldsymbol z}})f({{\boldsymbol y}})f({{\boldsymbol y}}+{{\bf h}}^{1/2}{{\boldsymbol z } } ) d{{\boldsymbol y}}d{{\boldsymbol z}}\sim r(f)\int\boldsymbol\varphi({{\boldsymbol z}})\boldsymbol\varphi({{\boldsymbol z}})^\top d{{\boldsymbol z}}\ ] ] which , in view of lemma [ lem : varphi ] , leads to @xmath229 . 1 . @xmath60 is a symmetric @xmath3-variate density such that @xmath232 and all its partial derivatives up to order @xmath233 are bounded , continuous and square integrable . 2 . @xmath234 is a sequence of bandwidth matrices such that all entries of @xmath235 and @xmath85 tend to zero as @xmath166 . to make use of lemma [ lem : asymhr ] once more , notice that the difference between the mise and its estimate is @xmath236 so taking into account @xmath220 again , we come to @xmath237&\sim(-1)^r \tfrac{m_2(k)^2}{2}(\operatorname{vec}^\top { { \bf i}}_{d^r}\otimes { { \bf i}}_{d^2 } \otimes \operatorname{vec}^\top { { \bf h}})(\hat{\boldsymbol{\psi}}_{2r+4}({{\bf g } } ) - \boldsymbol{\psi}_{2r+4}),\end{aligned}\ ] ] so that the performance of @xmath238 is determined by the performance of @xmath239 as an estimator of @xmath240 . from theorem 2 in @xcite the optimal pilot bandwidth @xmath85 for the estimator @xmath239 is of order @xmath130 , leading to @xmath241=o(n^{-4/(d+2r+6)})$ ] , and then @xmath242=o_p(n^{-2/(d+2r+6)}{\mathbf{j}}_{d^2})\operatorname{vec}{{\bf h}}.$ ] so finally we arrive to @xmath243 by applying lemma [ lem : asymhr ] . as in @xcite , it can be shown that the function @xmath244 can be replaced for @xmath245 everywhere in the asymptotic analysis , since the difference between their respective minimizers is of relative order faster than @xmath246 , which is the fastest attainable rate in bandwidth selection @xcite . so to apply lemma [ lem : asymhr ] it is also possible consider @xmath245 instead of @xmath244 , hence we focus on analyzing the difference @xmath247 at @xmath22 of the same order as @xmath116 . to begin with , note that using a fourth order taylor expansion of @xmath248 results in @xmath249 { \mathsf{d}}^{\otimes 2r+p } \bar{l}({{\bf g}}^{-1/2}{{\boldsymbol x } } ) \ , d{{\boldsymbol z}}\\ & = \tfrac{1}{4 } m_2(k)^2 \|{{\bf g}}\|^{-1/2 } ( { { \bf g}}^{-1/2})^{\otimes 2r } [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ( { { \bf g}}^{-1/2})^{\otimes 4 } ] { \mathsf{d}}^{\otimes 2r+4 } \bar{l}({{\bf g}}^{-1/2}{{\boldsymbol x}})\\ & = \tfrac{1}{4 } m_2(k)^2\|{{\bf g}}\|^{-1/2 } [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ] ( { { \bf g}}^{-1/2})^{\otimes ( 2r+4 ) } { \mathsf{d}}^ { \otimes 2r+4 } \bar{l}({{\bf g}}^{-1/2}{{\boldsymbol x}})\\ & = \tfrac{1}{4 } m_2(k)^2 [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ] { \mathsf{d}}^ { \otimes 2r+4 } \bar{l}_{{\bf g}}({{\boldsymbol x}}),\end{aligned}\ ] ] where we have made use of the fact that @xmath250 and @xmath251 , and that the entries of @xmath252 tend to zero as a consequence of ( a3 ) and ( a5 ) . this asymptotic approximation is then used to expand the terms in @xmath253= ( -1)^r { \operatorname{vec}^\top}{{\bf i}}_{d^r } \bigg\ { n^{-1 } \bar{\delta}_{{\bf h}}{\mathsf{d}}^{\otimes 2r } \bar{l}_{{\bf g}}(0 ) \\ + ( 1-n^{-1})\operatorname{\boldsymbol{\mathbb{e}}}\big [ ( \bar{\delta}_{{\bf h } } * { \mathsf{d}}^{\otimes 2r } \bar{l}_{{\bf g}})({{\bf x}}_1 - { { \bf x}}_2)\big ] - \int \bar{\delta}_{{\bf h } } * { \mathsf{d}}^{\otimes 2r } f({{\boldsymbol x } } ) f({{\boldsymbol x } } ) \ , d{{\boldsymbol x}}\bigg\}.\end{gathered}\ ] ] precisely , for the first term we have @xmath254({{\bf g}}^{-1/2})^{\otimes 2r+4 } { \mathsf{d}}^{\otimes 2r+4 } \bar{l}(0),\end{aligned}\ ] ] and for the second term @xmath255 \\ & \sim \tfrac{1}{4}m_2(k)^2[{{\bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ] \iint { \mathsf{d}}^{\otimes 2r+4 } \bar{l}_{{\bf g}}({{\boldsymbol x}}- { { \boldsymbol y } } ) f({{\boldsymbol x } } ) f({{\boldsymbol y } } ) \ , d{{\boldsymbol x}}d{{\boldsymbol y}}\\ & = \tfrac{1}{4}m_2(k)^2 [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ] \iint \bar{l}_{{\bf g}}({{\boldsymbol x}}- { { \boldsymbol y } } ) { \mathsf{d}}^{\otimes 2r+4}f({{\boldsymbol x } } ) f({{\boldsymbol y } } ) \ , d{{\boldsymbol x}}d{{\boldsymbol y}}\\ & \sim \tfrac{1}{4}m_2(k)^2 [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ] \iint \bar{l } ( { { \boldsymbol w } } ) \sum_{p=0}^2 \frac{(-1)^p}{p!}[{{\bf i}}_{d^{2r+4 } } \otimes ( { { \boldsymbol w}}^\top { { \bf g}}^{1/2})^{\otimes p } ] \\ & \quad \times { \mathsf{d}}^{\otimes 2r+4+p } f({{\boldsymbol y } } ) f({{\boldsymbol y } } ) \ , d{{\boldsymbol w}}d{{\boldsymbol y}}\\ & = \tfrac{1}{4}m_2(k)^2 [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } ] \sum_{p=0}^2 \frac{(-1)^p}{p!}[{{\bf i}}_{d^{2r+4 } } \otimes \{{{\boldsymbol \mu}}_p(\bar{l})^\top ( { { \bf g}}^{1/2})^{\otimes p}\}]{{\boldsymbol \psi}}_{2r+4+p}\\ & = \tfrac{1}{4}m_2(k)^2 [ { { \bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2}]{{\boldsymbol \psi}}_{2r+4 } + \tfrac{1}{4}m_2(k)^2 m_2(l)[{{\bf i}}_{d^{2r } } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } \otimes { \operatorname{vec}}^\top { { \bf g } } ] { { \boldsymbol \psi}}_{2r+6},\end{aligned}\ ] ] since @xmath256 and @xmath257 . finally , noting that @xmath258 and making use of the previously obtained expansion for @xmath259 , the third term is @xmath260 { { \boldsymbol \psi}}_{2r+4}.\end{aligned}\ ] ] thus , @xmath261 & \sim \tfrac{1}{4 } m_2(k)^2 n^{-1}\|{{\bf g}}\|^{-1/2}[{\operatorname{vec}}^\top{{\bf i}}_{d^r } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2}]({{\bf g}}^{-1/2})^{\otimes 2r+4 } { \mathsf{d}}^{\otimes 2r+4 } \bar{l}(0 ) \\ & \quad + \tfrac{1}{4}m_2(k)^2 m_2(l)[{\operatorname{vec}}^\top{{\bf i}}_{d^r } \otimes ( { \operatorname{vec}^\top}{{\bf h}})^{\otimes 2 } \otimes { \operatorname{vec}}^\top { { \bf g } } ] { { \boldsymbol \psi}}_{2r+6}\end{aligned}\ ] ] calculations in section [ sec:3 ] give @xmath85 is order @xmath262 , as for the plug - in selector , so substituting to this into the derivative of the previous equation yields @xmath263 \ } & = o ( [ n^{-1 } \|{{\bf g}}\|^{-1/2}(\operatorname{tr}{{\bf g}})^{-r-2 } + \operatorname{tr}{{\bf g } } ] { \mathbf{j}}_{d^2 } ) { \operatorname{vec}}{{\bf h}}\\ & = o(n^{-2/(2r+d+6 ) } { \mathbf{j}}_{d^2 } ) { \operatorname{vec}}{{\bf h}}.\end{aligned}\ ] ] lemma [ lem : asymhr ] shows that @xmath264 is asymptotically equivalent to @xmath265({{\bf h}}_{\mise , r})$ ] . since it was stated in section [ sec:3 ] that @xmath266 $ ] is dominated by its squared bias term , then @xmath267 . forina m. , armanino c. , lanteri s. and tiscornia e. ( 1983 ) classification of olive oils from their fatty acid composition . in : h. martens and h. j. russwurm ( eds . ) , _ food research and data analysis _ , applied science publishers , london , pp . 189214 . horton , p. and nakai , k. ( 1996 ) a probabilistic classification system for predicting the cellular localization sites of proteins . proceedings of _ intelligent systems in molecular biology ( ismb-96 ) _ , 109115 . zeng , q.t . , pratt , j.p . , pak , j. , ravnic , d. , huss , h. and mentzer , s.j . ( 2007 ) feature - guided clustering of multi - dimensional flow cytometry datasets . _ journal of biomedical informatics _ , * 40 * , 325331 .	important information concerning a multivariate data set , such as clusters and modal regions , is contained in the derivatives of the probability density function . despite this importance , nonparametric estimation of higher order derivatives of the density functions have received only relatively scant attention . kernel estimators of density functions are widely used as they exhibit excellent theoretical and practical properties , though their generalization to density derivatives has progressed more slowly due to the mathematical intractabilities encountered in the crucial problem of bandwidth ( or smoothing parameter ) selection . this paper presents the first fully automatic , data - based bandwidth selectors for multivariate kernel density derivative estimators . this is achieved by synthesizing recent advances in matrix analytic theory which allow mathematically and computationally tractable representations of higher order derivatives of multivariate vector valued functions . the theoretical asymptotic properties as well as the finite sample behaviour of the proposed selectors are studied . in addition , we explore in detail the applications of the new data - driven methods for two other statistical problems : clustering and bump hunting . the introduced techniques are combined with the mean shift algorithm to develop novel automatic , nonparametric clustering procedures which are shown to outperform mixture - model cluster analysis and other recent nonparametric approaches in practice . furthermore , the advantage of the use of smoothing parameters designed for density derivative estimation for feature significance analysis for bump hunting is illustrated with a real data example . _ keywords : _ adjusted rand index , cross validation , feature significance , nonparametric kernel method , mean integrated squared error , mean shift algorithm , plug - in choice
the discovery of several extra solar planets and brown dwarf candidates around solar type stars ( e.g. mayor & queloz 1995 , butler & marcy 1996 , cochran & hatzes 1997 , rebolo et al . 1995 , 1996 , basri et al . 1996 ) has advanced our understanding of the formation of planetary systems , but at the same time has also raised new astrophysical problems . notably , some of these planets are found to orbit very close to the central star ( e.g. mayor et al . 1997 ) and their orbits may be tidally unstable . for example , rasio et al . ( 1996 ) showed that the orbital decay of 51 peg s companion is inevitable ( although on a long timescale ) , and the authors conclude that the planet will ultimately spiral - in into the star . furthermore , the engulfing of planets is certainly a possibility when the star evolves to become a giant ( e.g. livio & soker 1984 ) . the present paper deals specifically with the reaction of a giant star `` swallowing '' a massive planet or a brown dwarf . this work was also motivated by the fact that the presence of a binary companion ( even of a substellar mass ) has often been invoked to explain the axisymmetrical morphology of elliptical and bipolar planetary nebulae ( pne ) ( e.g. iben & livio 1993 ; soker 1992 , 1996a ; livio 1997 ) . notably , if the substellar secondary is not too close to the primary , it will interact with the primary only on the upper agb ( e.g. soker 1995 , 1996b ) . in the course of its evolution , a main sequence star climbs along the red giant branch and eventually ( at a later stage ) can also reach the asymptotic giant branch ( agb ) . during the giant evolutionary phases , a planet may accrete from the wind emitted by the giant or be entirely engulfed by the giant s envelope . depending on the initial orbital separation , the planet or brown dwarf may be swallowed during the red giant phase , along the agb , or survive . once inside the envelope , due to viscous friction and also to tidal forces , a fraction of the orbital kinetic energy of the planet is converted into thermal energy and , as a consequence , the binary separation decreases . during this process , orbital angular momentum is imparted to the envelope . the spiraling - in of a planet or a brown dwarf inside the envelope of an agb star has been studied in several papers ( e.g. livio & soker 1984 ; soker , harpaz & livio 1984 ) . using simplifying assumptions , these authors were able to follow the evolution of the brown dwarf s mass subject to accretion and mass loss ( evaporation ) . their model reveals the existence of a critical mass @xmath6 below which the planet is evaporated or collides with the core . for the particular agb star model used , they found @xmath7 , but these results must be taken with caution since the physical processes involved in these calculations were treated only approximately . on the observational side , the orbital separations between close binary stars seem to indicate that low - mass secondaries tend to have small orbital separations ( e.g. tables 2 and 3 in iben & livio 1993 ) . a crude extrapolation of these data to low - mass stars ( soker 1996b ) suggests that secondaries with masses of @xmath8 will end up at orbital separations of @xmath9 . as we will see in the next section , it is likely that at these depths inside giants stars , the secondary will be tidally disrupted or evaporated . consequently , very low - mass secondaries will probably be accreted onto the agb star s core . in the present paper we follow the evolution of an agb star that accretes matter from the destruction of a brown dwarf or a massive planet , using a spherically symmetric stellar evolution code . in the next section , we present the initial model for the agb star and the general input physics . in section 3 , we describe the numerical approach and in section 4 we analyze the results of our computations . the effects of accretion on the nucleosynthesis and surface abundances are presented in section [ nucleo ] and a general discussion ( section [ discuss ] ) and summary follow ( section [ summary ] ) . to determine the exact location of the accretion process , we require a knowledge of the structure of the agb star and the colliding brown dwarf . where the brown dwarf will disrupt and dissipate can be estimated from the properties of the two objects involved in this scenario . in this section we thus present our initial model for the agb star and look into the properties of the brown dwarf . with these data at hand , we will try to characterize the accretion process as well as different quantities associated with this phenomenon . the initial agb star model ( taken from forestini & charbonnel 1997 ) has an age of @xmath10yr . its initial radius , mass , effective temperature and luminosity are equal to @xmath11 , @xmath12 ( initially @xmath13 ) , @xmath14k and @xmath15 , respectively . the star has already undergone 2 thermal pulses and is in the process of relaxing . the internal structure is composed of a very extended convective envelope which accounts for more than 99% of the radius . below the massive convective zone ( @xmath16 ) , we find a thin radiative shell surrounding the hydrogen burning shell ( hbs ) . in the hbs , h is mainly burned through the cno cycle . below the hbs is the intershell where h is exhausted but where the temperature is too low to allow for helium burning . below the intershell , helium is efficiently burned through 3@xmath17 reactions in the so - called helium burning shell ( hebs ) . at the centre of the star , below the hebs , the degenerate core is composed mainly of @xmath18o and @xmath19c . the core is highly degenerate and nearly isothermal . in table [ tab1 ] we summarize the properties of the agb star . note that the maximum temperature occurs off - centre due to neutrino losses . .physical properties of the initial agb model [ cols="<,^,^,^,^,^,^,^ " , ] a straightforward comparison of the values in tables [ tab1 ] and [ tab2 ] indicates that the brown dwarf s mean density is of the order of the density encountered in the hbs . it also reveals that brown dwarf radii are of the order of the radius of the hydrogen burning shell . given the properties of the agb star and the brown dwarf , we can now attempt to determine the locus of the brown dwarf dissipation and the ensuing accretion process . first , we estimate the virial temperature of the brown dwarf , since this represents the temperature above which the thermal kinetic energy of the star can overwhelm the gravitational binding energy of the brown dwarf . this is given by @xmath20 such temperatures , of the order of a few millions degrees , are encountered at the base of the convective envelope ( see table [ tab1 ] ) . therefore , initially relatively massive brown dwarfs ( @xmath21 ) will probably be able to reach the bottom of the convective envelope without being entirely evaporated . note that we do not expect the trajectory of the spiraling - in brown dwarf to be strongly affected by the convective motions since its keplerian velocity is always supersonic ( see e.g. livio & soker 1984 ) . as the brown dwarf gets closer to the stellar core , tidal effects become important and they can induce strong distortions to the brown dwarf s structure , and eventually they can even lead to a total disruption of the brown dwarf . the elongation stress at the brown dwarf centre can be approximated by ( e.g. soker et al . 1987 ) @xmath22 @xmath23 where @xmath24 is the mean density of the brown dwarf , @xmath25 and @xmath26 the mass and radius of the star causing the disruption , and @xmath27 is a parameter of order unity . if we compare this expression with the central pressure of the brown dwarf , then the condition of disruption , @xmath28 , reads @xmath29 @xmath30 from condition ( [ critere ] ) , it is clear that a disruption of the brown dwarf occurs in regions close to the bottom of the convective envelope . thus , all the indications are that the brown dwarf will be dissipated ( and its material accreted ) near the bottom of the convective layer . in order to estimate the mean accretion rate onto the agb star s core , we evaluate the orbital decay rate @xmath31 in the accretion region . the spiraling - in of the brown dwarf is due to the action of drag forces that result from viscous friction and tidal effects . this characteristic timescale , which is evaluated in the vicinity of the dissipation zone , provides a rough estimates of the transit timescale of the brown dwarf in this region and the rate at which matter can be deposited there . the accretion time scale is therefore given approximately by ( see e.g. livio & soker 1984 for a detailed discussion ) @xmath32 where @xmath33 is the density in the envelope at radial distance @xmath34 . consequently , the expected accretion rate is of the order of @xmath35for a 0.01 brown dwarf . the previous analysis indicates that a brown dwarf orbiting an agb star can spiral - in to very deep layers inside the star . on the basis of our estimates of the virial temperature and of tidal effects , we found that the brown dwarf is likely to be dissipated at the base of the convective envelope . the net outcome of the spiraling - in process is therefore the accretion of a brown dwarf mass at rates in the range of @xmath36 to @xmath37 . the stellar evolution code used in the present calculations has been previously used for the computation of grids of standard pre - main sequence ( pms ) evolutionary tracks ( forestini 1994 , siess et al . 1997a ) , for the detailed evolution and nucleosynthesis of intermediate mass agb stars ( forestini & charbonnel 1997 ) , and also for the effects of disc accretion onto pms stars ( siess et al . 1997b , 1998 ) . we refer the reader to these papers for a detailed description of the stellar evolution code . we computed two sets of models for two different accretion rates , @xmath38 and @xmath36 ( models a and b , respectively ) . the total accreted mass in cases a and b was 0.01245 and 0.01238 , respectively . the chemical composition of the accreted matter was assumed to correspond to a solar metallicity ( @xmath39 ) with relative abundances from anders and grevesse ( 1989 ) . previous works dealing with accretion in different evolutionary phases ( e.g. stahler 1988 , siess et al . 1997b ) have pointed out the important role of deuterium in this process . indeed , the very fast burning of h through the h(p,@xmath40)he reaction releases a huge amount of energy which is able to modify the structure and evolution of the star significantly . the presence or absence of deuterium in the accreted matter is thus an important issue which depends essentially on the mass of the brown dwarf . brown dwarfs ( or planets ) more massive than @xmath41 jupiter masses ( @xmath42 , e.g. burrows et al . 1995 ) have a high enough central temperature to burn deuterium in their interiors . consequently , we do not expect deuterium to be accreted in the agb star and its initial mass fraction has been set to be @xmath43 . we also assume that li is preserved in the brown dwarf and its mass fraction ( abundance ) is equal to @xmath44 [ log@xmath45(li)=3.31 ] . as we discussed in section [ sect ] , a sufficiently massive brown dwarf is able to reach the bottom of the convective envelope . the deposition of some accreted matter inside the convective envelope , due to partial evaporation , will probably not in itself influence the structure since this region is devoid of energy sources . the main effects that need to be studied , therefore , are the consequences of mass deposition in the region close to the bottom of the convective zone , where the planet / brown dwarf is expected to disrupt . to achieve this goal , we have assumed a simplified mass deposition profile in which the mass @xmath46 of the shell @xmath47 is increased at every time step @xmath48 by the amount @xmath49 , where @xmath50 . the mass is accreted in the region delineated by @xmath51 , where @xmath52 corresponds to the mass coordinate at the top of the hbs . however , if hot bottom burning is taking place ( see below ) , @xmath52 is equal to the bottom of the convective envelope ( which is inside the hbs ) . with this prescription , we allow the accreted matter to go below the convective envelope , but , as we shall see later , a small fraction of the accreted mass ( @xmath53 ) is actually deposited in this region . the boundary @xmath54 is defined ( arbitrarily ) as the mass coordinate corresponding to a radius equal to 1(@xmath55 ) . this accounts for the fact that the planet is actually dissipated in the central region of the star . we also assume that the accreted matter is injected at the local envelope s temperature and density . this can be justified by the fact that the brown dwarf s kinetic energy is not too different from the specific thermal energy of the agb envelope at the same point ( e.g. harpaz & soker 1994 ) . + in the presence of mass accretion , at each evolutionary time - step , the total mass of the star is increased by the amount @xmath56 . in the shells where the mass is deposited ( for @xmath51 ) , the time derivative of a variable @xmath57 is now calculated as @xmath58 where @xmath59 is the velocity of the shell and @xmath60 . the rate of gravitational energy release , @xmath61 , is given by @xmath62 where @xmath63 is the specific internal energy . the chemical composition entering the computation at the new time step is modified due to the arrival of fresh , not nuclearly processed material from the accreted matter . the mass fraction @xmath64 of element @xmath65 at shell @xmath47 after mixing , is given by ( siess et al . 1997b ) @xmath66 where @xmath67 is the stellar mass fraction of element @xmath65 before mixing and @xmath68 is the mass fraction this element in the accreted matter . note that @xmath68 is constant and does not depend on the mass coordinate . finally , in order to speed - up the convergence process , we turned off the wind mass loss during the accretion phase and we used grey atmosphere models . the evolution of the agb star under consideration can be divided into 2 major phases . the first one deals with the onset of accretion and the main accretion phase while the second marks the end of the accretion process and the relaxation of the star . the deposition of material inside the star releases a significant amount of gravitational energy . the associated accretion luminosity is given by @xmath69 note that an additional source of energy can arise from the possible replenishment of the nuclearly active region with accreted material . the modification of the chemical composition of these regions can increase the nuclear energy production . for high accretion rates ( case a ) , represents a large fraction of the stellar luminosity and in the deep region where the energy is deposited , the star expands . locally , the density , pressure and temperature decrease . the hbs , which advances in mass , rapidly reaches the `` cooler '' accretion region where the mass is deposited . note also that the top of the accretion region is located in the convective envelope . ] and then , the nuclear energy production due to h burning decreases ( see fig . [ my404_01 ] ) . on the other hand , the luminosity of the hebs increases , since the addition of mass results in an increased gravity and compressional heating . in the expanding accretion region , the opacity @xmath70 increases since in this regime it is a decreasing function of the temperature . the increase in both @xmath70 and @xmath71 ( due to ) contributes to a steepening of the radiative gradient ( @xmath72 ) which now becomes greater than the adiabatic gradient , thus convection sets in . consequently , the convective envelope penetrates rapidly towards deeper regions ( see fig . [ my404_02 ] ) and reaches the hbs , leading to the so - called hot bottom burning ( hbb ) . with the onset of hbb , the luminosity of the hbs ( @xmath73 ) rises sharply , as fresh material coming from the convective region is engulfed in this shell . the global contribution to the total nuclear energy production of the c(p,@xmath40)n reaction is increased during this phase , and it accounts for 27% ( compared to 22% in a standard evolution ) . we also note that the n(p,@xmath40)o and n(p,@xmath17)o reactions contribute less to the nuclear energy production than in a standard scheme . these different features indicate that the cno cycle is reaching a new equilibrium . the `` flash '' induced by the penetration of the convective envelope into the hbs increases @xmath73 by 4 orders of magnitudes . the large release of nuclear energy is used to expand the envelope and shortly after the ignition of the hbb , @xmath73 decreases rapidly ( fig . [ my404_01 ] ) . thereafter , provides most of the luminosity , the hbs almost extinguishes as it moves into the colder accretion region , whereas @xmath74 rises since the mass increases . figure [ my404_03 ] depicts the profiles of several physical quantities during the early accretion phase . the peaks in the luminosity profiles indicate that the star is not in thermal equilibrium , but rather energy accumulates in its interior . one also notices that this feature is smoothed out as time advances , as a result of the star s expansion . in our initial model ( solid line ) , the gravitational energy production @xmath61 represents a small fraction of the stellar energy production , 80% of the luminosity is due to nuclear energy . when the accretion process starts ( dotted line ) , @xmath61 increases rapidly below the accretion region as a result of mass deposition . in the accretion region where @xmath75 , the expansion induces modifications to the temperature profile which is now steeper . the nuclear energy production is consequently more localized and concentrated to thinner regions . when hbb starts ( short - dashed line ) , both @xmath61 and @xmath76 reach extremum values . the luminosity becomes negative , as most of the energy production is used to expand the star . finally , when the hbb is settled ( long - dashed line ) , the star heats up in the region of nuclear energy production ( @xmath77 ) , whereas expansion proceeds above this zone . during the accretion phase the hbs narrows as the temperature profile becomes steeper ( fig . [ my404_02 ] ) . this also accounts for the global decrease of @xmath78 over this period . finally , since mass is being added , all the structural features ( location of the maximum energy production , burning shells , step in temperature ) move towards higher mass coordinates . in case b , the release of accretion energy is not sufficient to expand the star and trigger convective energy transport down to the hbs . consequently , a thin radiative layer is maintained below the convective envelope , like in the standard case ( fig . [ my404_02 ] ) . this thin layer prevents the energy production of the hbs from heating the envelope directly and , as is small and the gravitational pull increases , the star contracts . at the beginning of the accretion process , the release of gravitational energy lowers the temperature in the accretion region ( above the hbs ) . as the core grows due to mass addition , the hbs moves outwards and rapidly penetrates the less dense and cooler accretion region . as a result , the nuclear luminosity due to hydrogen burning ( @xmath73 ) undergoes a sudden drop ( fig . [ my404_04 ] ) . the star compensates for this decrease in energy production by a small contraction . thereafter , the star tries to find an equilibrium configuration in which @xmath78 and account for the total luminosity . this phase of structural readjustment leads to a small expansion . thermal equilibrium is finally achieved after @xmath79yr , which corresponds to the stellar kelvin helmholtz timescale @xmath80 subsequently , contraction resumes and the nuclear luminosity stays at an almost constant value , somewhat lower than in the standard case ( without accretion ) , due to the presence of . however , the increase in mass speeds - up the contraction rate in the central region and the luminosity of the hebs rises . + the main difference between case b and the standard evolution is the persistence of stellar contraction , not only in the core but throughout the entire structure during most of the accretion phase . for low accretion rates , it is mainly the mechanical effects of mass deposition and gravitational attraction that prevail . for a comparison with the high accretion rate case , we present in fig . [ my404_05 ] the profiles of several physical variables during this period . as can be seen , @xmath61 remains largely positive in the accretion region : contraction heats the central parts of the star . the profiles of @xmath81 , @xmath76 and @xmath71 are very similar during the entire phase . we do not obtain a significant increase in the steepness of the temperature gradient nor a peak in the luminosity profile . the different quantities are simply shifted to higher mass coordinates very smoothly . in this case , accretion does not lead to significant structural changes , the energetic effects of accretion ( ) being too weak . for high accretion rates ( fig . [ my404_01 ] ) , the sudden suppression of accretion energy deposition initiates a rapid contraction of the star , as the envelope can no longer be supported . the release of gravitational energy increases the temperature in the regions of energy production and induces an enhancement of the nuclear luminosities . the star globally heats up , the opacity drops , and the convective envelope recedes towards the surface ( fig . [ my404_02 ] ) . the cessation of the accretion process thus marks the end of hbb and this leads to the formation a thin radiative layer between the convective zone and the hbs . the contraction proceeds as long as the energy losses from the surface are not equilibrated by the gain from gravitational and nuclear energy sources . the return to equilibrium is finally achieved after @xmath82yr , when the stellar radius and luminosity have decreased by a factor of @xmath83 ( fig . [ my404_06 ] ) . at this point , the nuclear luminosity provides more than 90% of the stellar luminosity and it is sufficient to support the envelope . as most of the nuclear energy is produced by the hbs , the energetic influence of the hebs is shielded and its luminosity ( @xmath74 ) decreases ( as in the standard case ) . finally , as the core mass @xmath84 grows , the temperature inside the hebs increases and the nuclear reactions ( which are very temperature sensitive ) runaway and trigger a thermal instability . note that in case a , the thermal pulse is delayed with respect to the standard model . this is due to the large expansion that occurs during the accretion phase , which in turn slows down the `` natural '' temperature increase of the hebs substantially . in case b , when the accretion process stops , the star undergoes a short phase of rapid contraction with the disappearance of . the temperature in the hbs increases and the nuclear luminosity rises ( fig . [ my404_06 ] ) . for a short period of time the star expands . finally , when an equilibrium configuration is reached , which accounts for the missing , the star resumes contraction and @xmath74 keeps rising . the tendency of increasing @xmath74 is due to the fact that for low accretion rates the impact of accretion is more a consequence of the mechanical effects resulting from mass deposition ( compressional heating ) , rather than energetic effects due to or hbb . note in addition that the accretion rate used in this study is larger that the `` normal '' core growth rate ( @xmath85 , for a 3 star ) . therefore , accretion forces the star to evolve more rapidly than in a standard evolution and the thermal pulse occurs earlier . however , the thermal instability is weaker in the star accreting the brown dwarf than in the standard model : the maximum of @xmath74 is more than an order of magnitude lower than in the conventional computation . the reason for this difference is the fact that due to the mass deposition , the hebs moves towards regions of a lower density faster than in the standard scheme , and the nuclear energy generation is thus reduced . if a high accretion rate ( @xmath2 ) is sustained for a longer period of time , the release of gravitational energy maintains the envelope expansion . after @xmath86yr , the kelvin helmholtz timescale has decreased to @xmath87yr and accretion energy can now be efficiently radiated away . at that time , the nuclear energy production is at a minimum and the hbs is almost extinguished ( fig . [ my404_07 ] ) . the luminosity profile is now monotonicly increasing and contraction resumes . when the accretion process ends ( for an accreted mass @xmath88 in this simulation ) , the contraction rate accelerates as the envelope is no longer supported by the deposition of accretion energy . the temperature increases and @xmath74 rises first , followed by @xmath73 . when the nuclear sources provide @xmath89% of the stellar luminosity the contraction stops . the radius is then minimal and its value is smaller than in case a , because in this simulation , the stellar mass and thus the gravitational pull are larger . finally , as the temperature increases the thermal instability sets in , but somewhat later than in case a. for lower accretion rates ( @xmath90 ) , the persistent addition of mass , up to @xmath91 , increases the contraction rate . the star heats up faster and the thermal instability develops even earlier than in case b ( fig . [ my404_08 ] ) . for the reasons explained previously ( see section [ sect1 ] ) , the value of @xmath74 during the pulse is lower than in the standard case . we also note that accretion does not perturb the thermal pulse , which indicates that in this case this process represents a small perturbation to the stellar structure . when the instability is over , the star resumes contraction . the nuclear luminosity ( @xmath78 ) is lower than in the standard model because a fraction of the energy is provided by the accretion process . finally , when accretion ends , the star relaxes and returns to a standard evolution ( with a larger mass however ) . to summarize these simulations , for high accretion rates the accretion luminosity dominates the stellar energetics , reducing the nuclear energy production of the hbs . as a result , the ( nuclear ) evolution of the star is slowed down . for lower accretion rates ( but still high compared to the normal core growth rate ) , accretion shifts the internal structure to that corresponding to higher masses at a faster rate than in a standard evolution , and consequently the evolution is accelerated . however , these results do not imply that for some intermediate value of the evolution of the star remains unchanged . indeed , the stellar mass increases and the core s growth rate remains always higher than in a standard evolution . in the limiting case where the release of accretion energy exactly counterbalances the gravitational attraction , instantaneously the evolution of the star would remain the same . however this situation is unstable since @xmath84 increases . quite rapidly the gravitational pull takes over and the contraction resumes . note that for large accretion rates when the accretion process is maintained for a longer period of time , contraction finally resumes once thermal equilibrium is reached . chemical modifications resulting from mass accretion are expected if either a substantial amount of accreted matter is deposited in the envelope or / and if nuclear burning is operating in the convective envelope . to separate these two effects , we define new variables @xmath92 and @xmath93 for each chemical element @xmath65 by @xmath94 where @xmath95 is the mass fraction of element @xmath65 in an evolutionary scheme with accretion and @xmath96 is the mass fraction of this element in standard computations . @xmath97 represents the ratio of the total mass deposited in the envelope during the accretion process ( @xmath98 ) to the mass of this region ( @xmath99 ) . the variable @xmath92 describes the effect of accretion on the surface abundances , including mass deposition and nuclear burning . the variable @xmath93 represents the nuclear activity : it gives the ratio of the mass fraction computed in our models with accretion to the mass fraction expected from calculations with accretion if no nuclear burning took place . therefore , if the element is nuclearly produced then @xmath100 and if it is destroyed @xmath101 . the deposition of fresh , nuclearly unaltered material in the convective envelope entails modifications of the surface chemical abundances . this effect accounts for the large values of @xmath92 attached to light elements such as li , li , be , b or b ( fig . [ my404_09 ] ) . indeed , due to their low burning temperature , these elements were previously burned during the pre - main sequence phase or during the first dredge up . depending on the mass added to the envelope , the abundances of these elements can be increased by a factor of @xmath102 ( @xmath103 dex ) . the more mass is accreted , the higher the ratio @xmath92 is . however , the detection of these chemical changes may be difficult if the accreted mass is small . for example , the li mass abundance in cases a and b ( @xmath104 ) reaches values of log@xmath45(li)=1.54 and log@xmath45(li)=1.51 respectively , which are very close to the log@xmath45(li)=1.40 standard value . it is interesting to note that the accretion of planetary material in the envelope of red giant stars was originally suggested by alexander ( 1967 ) to explain the high li abundance observed in some red giants . brown et al . ( 1989 ) showed that a sufficient mass of terrestrial - like planets ( which have lost their initial h and he content ) may replenish the envelope of giants sufficiently to the reproduce high observed li abundances . note , finally , that the surface abundances of heavier elements , with @xmath105 , are close to their initial values and are thus not modified by the accretion process . the upper panel of fig . [ my404_09 ] indicates which elements undergo efficient nuclear reactions . the light elements are the principal target of nuclear burning . indeed the `` low '' temperature at the base of the convective region , between @xmath106 and @xmath107k ( depending on the presence or the absence of hbb ) , allows the destruction of these species . the nuclear burning acts preferentially on li , li and be . we can also detect the effects of cno reactions in the envelope through the production of c , n and o and the depletion of c. however , the influence of h burning through the cno cycle is very small and it only appears if accretion is maintained for a sufficiently long period of time ( dotted line , dashed line in fig . [ my404_09 ] ) . to summarize , the temperature at the base of the convective envelope ( @xmath108k ) is too low to give rise to a rich nucleosynthesis . the main chemical changes thus result from the deposition of fresh material in the convective envelope . our computations involved a relatively young agb star ( at the beginning of the thermally pulsing phase ) and one may wonder what would be the effects of brown dwarf accretion inside more evolved stars . in this respect we note , that as the evolution progresses , the core grows but the temperature at the base of the convective zone remains almost constant , close to @xmath109k . thus , the location of the accretion process may remain unchanged . however , in older agb stars the stellar luminosity is substantially higher and this will most certainly result in a different stellar response to high accretion rates . indeed , as we have seen in the present work , the main reason for the stellar expansion is the release of a large amount of accretion energy ( @xmath110 ) . we thus expect that if mass is injected in more luminous agb stars , then the stellar radius and luminosity will decrease during the process , as in case b ( see also harpaz & soker 1994 ) . our calculations revealed that for high mass injection rates ( @xmath38 ) the radius and luminosity of the star increase by a factor of @xmath83 . this leads to a substantial increase in the mass loss rate , since according to the reimers formula ( 1975 ) @xmath111 . the ejected mass can be estimated as @xmath112 where @xmath113 is the duration of the accretion process . we can therefore expect the formation of shells , corresponding to individual accretion events . the mass of these shells is of the order of @xmath114 . a phenomenon not directly addressed by our spherically symmetric calculations is that of the deposition of angular momentum into the giant s envelope , by the accreted brown dwarf . we can however estimate the effects of such a deposition as follows . if we assume that the secondary deposits all of its orbital angular momentum in the envelope , and we further assume that the envelope rotates rigidly ( due to the fact that it is convective , and thus viscous ) then the ratio of the envelope angular velocity @xmath115 to the critical velocity ( surface keplerian angular velocity ) @xmath116 is given by @xmath117 where @xmath118 is the initial orbital radius of the brown dwarf . here we calculated the moment of inertia of the envelope for our model , and it is given by @xmath119 , where @xmath120 . as we can see , eq . [ omega ] indicates that for typical brown dwarf masses , the agb star can be spun - up to a significant fraction of the critical velocity . notably , the accretion of angular momentum is required to explain the abnormally high rotational velocities of fk comae stars ( e.g. rucinski 1990 , welty & ramsey 1994 ) and of peculiar agb stars such as v hydrae ( barnbaum et al . finally , this can also have important consequences for the shaping of planetary nebulae ( see e.g. soker 1996a ) . in particular , it has been suggested that the spin - up of the envelope is responsible for the high fraction of non - spherical planetary nebulae ( see livio 1997 for a review ) . concerning peculiar abundances , the discovery of lithium rich g and k giant stars ( e.g. wallerstein & sneden 1982 , brown et al . 1989 , de la reza & da silva 1995 ) posed the problem of the origin of this element in these stars . indeed , standard stellar evolution predicts that li should be depleted during the pre - main sequence phase or during the first dredge up . several mechanisms have been proposed to explain the high li abundance , these include : ( 1 ) li production ( cameron & fowler 1971 ) , ( 2 ) the retainment of the original lithium , ( 3 ) the external enrichment by novae ( e.g. gratton & dantona 1989 ) , ( 4 ) the engulfing of a planet ( e.g. brown et al . 1989 ) . recently , de la reza et al . ( 1996 , 1997 ) pointed out that almost all the li k giants are optical counterparts of iras sources , the far ir excesses being attributed to the presence of a dusty circumstellar shell ( cs ) . based on this apparent correlation between the ir and li excesses , these authors proposed a scenario in which all the k giants with masses between 1.0 and 2.5 become li rich after the first dredge - up . they assume that the mechanism responsible for the li enrichment is accompanied by a sudden mass loss which leads to the formation of a cs . the li which has not been ejected remains in the stellar surface and is depleted later when the mass loss stops . in this scenario , the cs expands , cools down , disappears and the li surface abundance reaches again standard values . this evolutionary path , or `` li path '' , can reasonably explain the li abundance distribution of k giant stars in a colour - colour diagram . its duration is of the order of @xmath121yr and it requires that li suddenly appear at the surface of the star and then be depleted on a timescale of @xmath122yr . however , de la reza et al . do not provide convincing mechanisms for li enrichment and mass loss . they invoke an extra mixing process for li production , the `` cool bottom processing '' ( wasserburg , boothroyd & sackmann 1995 ) , but they remain vague concerning the generation of mass loss . we propose here that the accretion of a planet or a brown dwarf by a giant star could effectively explain all of these features . indeed , our calculations show that the accretion process ( 1 ) can increase the mass loss rate and lead to the ejection of a shell ( eq . [ shell ] ) , and ( 2 ) can enrich the envelope with lithium ( fig . [ my404_09 ] ) , in agreement with observations . observations also show that lithium rich giants have generally normal rotational velocities [ de medeiros et al . 1996 ; although some rapid rotators exist , e.g. 1e 1751 + 7046 ( ambruster et al . this implies that if the accretion of a planet / brown dwarf is responsible for the li enhancement , some fraction of the orbital angular momentum which is deposited into the envelope has to be removed ( or that in most cases of engulfed planets @xmath123 , see eq . [ omega ] ) . this can be achieved by efficient magnetic braking , since the giants develop a deep convective envelope ( e.g. leonard & livio 1995 ) . we should note that the paucity of giants showing both rapid rotation and high li abundance probably indicates that the timescale for rotational braking is short compared to the giant evolutionary phase ( a few @xmath107yr ) . the existence of very rich li stars ( e.g. gregorio - hetem et al . 1992 , torres et al . 1995 , de la reza & da silva 1995 , fekel et al . 1996 ) poses a somewhat more difficult challenge . one could imagine , however , that the collision of the planet ( or a brown dwarf ) with the star s core may trigger the `` cool bottom processing '' ( wasserburg , boothroyd & sackmann 1995 ) . in this scenario , be - rich material can be dredged - up from the deep interior to the convective envelope where it decays through the be(@xmath124)li reaction . a comment should be added about statistics . from a sample of 644 stars , brown et al . ( 1989 ) found that @xmath125% of g and k giants have li abundances above @xmath126 and that an additional 4% have abundances between 1.2 and 1.3 . considering the fact that many of the li rich stars show an ir excess compatible with the presence of a surrounding shell , it is tempting to interpret these observations as a signature of an accretion event . this could indicate that @xmath127% of all stars have low - mass ( possibly substellar ) companions . these statistics are fully compatible with those inferred from pne nuclei ( e.g. yungelson , tutukov & livio 1993 ) . we have presented computations in which a brown dwarf ( or a planet ) is accreted in the core of an agb star . this situation results when the brown dwarf is engulfed in the stellar envelope as the latter is expanding due to stellar evolution . viscous friction as well as tidal effects generate a drag force that makes the brown dwarf spiral in . we have shown that the region where the brown dwarf is expected to disrupt is close to the base of the convective envelope . we investigated the response of the star to the accretion of a small body for two accretion rates , @xmath37 and @xmath36 . our simulations show that for the higher accretion rate , the main energetic effects come from the release of potential energy . the accretion luminosity is of the order of the stellar luminosity and this additional energy source makes the star expand . another interesting feature of these computations ( involving a high accretion rate ) is the appearance of hot bottom burning . however , the relatively low temperature at the base of the convective envelope ( @xmath128k ) and the short duration of the accretion phase do not allow for a rich nucleosynthesis in the convective zone . we find that accretion delays the development of the thermal instability . the nuclear energy production due to the h and he burning shells is smaller than in a standard case because a part of the energy is provided by the deposition of the accretion energy . hydrogen and helium are consequently burned more slowly and the evolution is slowed down . for the lower accretion rate the evolution is different . in this case the accretion luminosity represents a small fraction of the stellar luminosity and therefore expansion does not ensue . conversely , the addition of mass increases the gravitational pull and the star contracts even more efficiently than in a standard scheme . the temperature of the nuclear burning shells increases more rapidly and the thermal instability occurs earlier than in conventional evolution . the global stellar evolution is accelerated . except for variations in the luminosity and radius , both of which can be increased by a factor of @xmath83 , the effects of mass accretion in an evolved star may be traced by changes in the surface chemical abundances . we showed that such changes are mainly the result of material deposition rather than nucleosynthesis . the chemical changes affect mainly light elements such as li and be whose mass fraction can be increased by a factor of 2 - 8 , depending on the accretion rate and duration of the accretion process . generally , the observational signatures of the accretion of a brown dwarf or a planet by giant stars can be ( depending on the physical parameters involved ) : @xmath129 the formation of a circumstellar shell , @xmath130 the enhancement of the surface li abundance , @xmath131 rapid rotation . we propose that the ir excesses and high li abundances observed in @xmath127% of the g and k giants originate from the accretion of massive planets / brown dwarfs / very low - mass stars . finally , we should note that we also made several attempts to accrete matter deeper inside the star ( e.g. below the hbs ) . however , such a process could not be followed reliably with our code . the deposition of protons in the very central regions increases the nuclear energy production to such rates , that the envelope can not absorb this energy excess . ls wishes to thank manuel forestini for invaluable discussions and also for kindly providing us with the initial stellar model . ls acknowledges support from the french ministry of foreign affairs ( bourse lavoisier ) and thanks the stsci for its hospitality . this work has been supported in part by nasa grants nagw-2678 , g005.52200 and g005.44000 , and the director s discretionary research fond at stsci . the computations presented in this paper were performed at the `` centre intensif de calcul de lobservatoire de grenoble '' . 99 alexander j.b . , 1967 , observatory , 87 , 238 ambruster c.w . , fekel f.c . , guinan e.f . , hrivnak b.j . , 1997 , apj , 479 , 960 anders e. , grevesse n. , 1989 , geochim . acta , 53 , 197 basri g. , marcy g.w . , graham j.r . , 1996 , apj , 458 , 600 barnbaum c. , morris m. , kahane c. , 1995 , apj , 450 , 862 burrows a. , hubbard w.b . , saumon d. , lumine j.i . , 1993 , apj , 406 , 158 burrows a. , saumon d. , guillot t. , hubbard w.b . , lumine j.i . , 1995 , nature , 375 , 299 butler r.p . , marcy g.w . , 1996 , apj , 464 , l153 brown j.a . , sneden c. , lambert d.l . , dutchover e.jr . , 1989 , apjs , 71 , 293 cameron a.g.w . , fowler w.a . , 1971 , apj , 164 , 11 cochran w.d . , hatzes a. , marcy g.w . , butler r.p . , 1997 , apj , 483 , 457 de la reza r. , da silva l. , 1995 , apj , 439 , 917 de la reza r. , drake n.a . , da silva l. , 1996 , apj , 456 , l115 de la reza r. , drake n.a . , da silva l. , torres c.a.o . , martin e.l . , 1997 , apj , 482 , l77 de medeiros j.r . , melo c.h.f . , mayor m. , 1996 , a&a , 309 , 465 fekel f.c . , webb r.a . , white r.j . , zuckerman b. , 1996 , apj , l95 forestini m. , 1994 , a&a , 285 , 473 forestini m. , charbonnel c. , 1997 , a&as , 123 , 241 gratton r.g . , dantona f. , 1989 , a&a , 215 , 66 gregorio - hetem j. , lepine j.r.d . , quast g.r . , torres c.a.o . , de la reza r. , 1992 , aj , 103 , 549 harpaz a. , soker n. , 1994 , mnras , 270 , 734 iben i.jr . , livio m. , 1993 , pasp , 105 , 1373 leonard p.t.j . , livio m. , 1995 , apj , 447 , l21 livio m. , 1997 , space sci . , in press livio m. , soker n. , 1984 , mnras , 208 , 763 mayor m. , queloz d. , 1995 , nature , 378 , 355 mayor m. , queloz d. , udry s. , halbwachs j .- l . , 1997 , in `` astronomical and biochemical origins and the search for life in the universe '' , proceedings of the 5th international conference on bioastronomy , iau coll . 161 , ed c. cosmovici , s. bowyer , d. werthimer , p313 rasio f.a . , tout c.a . , lubow s.h . , livio m. , 1996 , apj , 470 , 1187 rebolo r. , martin e.l . , basri g. , marcy g. , zapatero - osorio m.r . , 1996 , apj , 469 , l53 rebolo r. , zapatero - osorio m.r . , martin e.l . , nature , 1995 , 377 , 129 reimers d. , 1975 , men . lige , 6th ser . 8 , 369 rucinski s.m . , 1990 , pasp , 102 , 306 siess l. , forestini m. , dougados c. , 1997a , a&a , 324 , 556 siess l. , forestini m. , bertout c. , 1997b , a&a , 326 , 1001 siess l. , forestini m. , bertout c. , 1998 , a&a , in press soker n. , 1992 , apj , 389 , 628 soker n. , 1995 , mnras , 274 , 147 soker n. , 1996a , apj , 460 , l53 soker n. , 1996b , apj , 468 , 774 soker n. , harpaz a. , livio m. , 1984 , mnras , 210 , 189 soker n. , regev o. , livio m. , shara m.m . , 1987 , apj , 318 , 760 stahler s.w . , 1988 , apj , 332 , 804 stevenson d.j . , 1991 , araa , 29 , 163 torres c.a.o . , quast g.r . , de la reza r. , gregorio - hetem j. , lepine j.d . , 1995 , aj , 109 , 2146 wallerstein g. , sneden c. , 1982 , apj , 225 , 577 wasserburg g.j . , boothroyd a.i . , sackmann i.j . , 1995 , apj , 447 , l37 welty a.d . , ramsey l.w . , 1994 , apj , 435 , 848 yungelson l.r . , tutukov a.v . , livio m. , 1993 , apj , 418 , 794 zapolsky h.s . , salpeter e.e . , 1969 , apj , 158 , 809	we study the response of the structure of an asymptotic giant branch ( agb ) star to the accretion of a brown dwarf or planet in its interior . in particular , we examine the case in which the brown dwarf spirals - in , and the accreted matter is deposited at the base of the convective envelope and in the thin radiative shell surrounding the hydrogen burning shell . in our spherically symmetric simulations , we explore the effects of different accretion rates and we follow two scenarios in which the amounts of injected mass are equal to @xmath0 and @xmath1 . the calculations show that for high accretion rates ( @xmath2 ) , the considerable release of accretion energy produces a substantial expansion of the star and gives rise to hot bottom burning at the base of the convective envelope . for somewhat lower accretion rates ( @xmath3 ) , the accretion luminosity represents only a small fraction of the stellar luminosity , and as a result of the increase in mass ( and concomitantly of the gravitational force ) , the star contracts . our simulations also indicate that the triggering of thermal pulses is delayed ( accelerated ) if mass is injected at a slower ( faster ) rate . we analyze the effects of this accretion process on the surface chemical abundances and show that chemical modifications are mainly the result of deposition of fresh material rather than of active nucleosynthesis . finally , we suggest that the accretion of brown dwarfs and planets can induce the ejection of shells around giant stars , increase their surface lithium abundance and lead to significant spin - up . the combination of these features is frequently observed among g and k giant stars . -1 cm 2@xmath4h # 1@xmath5 accretion planetary systems stars : agb and post - agb stars : evolution stars : low - mass , brown dwarfs stars : abundances
excitons in semiconductor crystals @xcite and nanostructures @xcite,@xcite are a very interesting and challenging object to search for the process of bose - einstein condensation ( bec ) . nowadays there is a lot of experimental evidence that the optically inactive para - excitons in cu@xmath0o can form a highly correlated state , or the excitonic bose einstein condensate @xcite,@xcite,@xcite . a moving condensate of para - excitons in a 3d cu@xmath0o crystal turns out to be spatially inhomogeneous in the direction of motion , and the registered velocities of coherent exciton packets turn out to be always less , but approximately equal to the longitudinal sound speed of the crystal @xcite . analyzing recent experimental @xcite,@xcite,@xcite and theoretical @xcite-@xcite studies of bec of excitons in cu@xmath1o , we can conclude that there are essentially two different stages of this process . the first stage is the kinetic one , with the characteristic time scale of @xmath2ns . at this stage , a condensate of long - living para - excitons begins to be formed from a quasi - equilibrium degenerate state of excitons ( @xmath3 , @xmath4 ) when the concentration and the effective temperature of excitons in a cloud meet the conditions of bose - einstein condensation @xcite . note that we do not discuss here the behavior of ortho - excitons ( with the lifetime @xmath5ns ) and their influence on the para - exciton condensation process . for more details about the ortho - excitons in cu@xmath1o , ortho - para - exciton conversion , etc . see @xcite,@xcite , @xcite , @xcite . the most intriguing feature of the kinetic stage is that formation of the para - exciton condensate and the process of momentum transfer to the para - exciton cloud are happening simultaneously . if the diameter of an excitation spot on the crystal surface is large enough , @xmath6 , and the energy of a laser beam satisfies @xmath7 , nonequilibrium acoustic phonons may play the key role in the process of momentum transfer . as a result , the mode with macroscopical occupancy of the excitons appears to be with @xmath8 where @xmath9 and @xmath10 is the packet velocity . indeed , the theoretical results obtained in the framework of the `` phonon wind '' model @xcite,@xcite and the experimental observations @xcite,@xcite,@xcite are the strong arguments in favor of this idea . to the authors knowledge , there are no realistic theoretical models of the kinetic stage of para - exciton condensate formation where quantum degeneracy of the appearing exciton state and possible coherence of nonequilibrium phonons pushing the excitons would be taken into account . indeed , the condensate formation and many other processes involving it are essentially nonliner ones . therefore , the condensate , or , better , the _ macroscopically occupied mode _ , can be different from @xmath11 , and the language of the states in @xmath12-space and their occupation numbers @xmath13 may be not relevant to the problem , see @xcite . in this study , we will not explore the stage of condensate formation . instead , we investigate the second , quasi - equilibrium stage , in which the condensate has already been formed and it moves through a crystal with some constant velocity and characteristic shape of the density profile . in theory , the time scale of this `` transport '' stage , @xmath14 , could be determined by the para - exciton lifetime ( @xmath15s @xcite ) . in practice , it is determined by the characteristic size @xmath16 of a high - quality single crystal available for experiments : @xmath17 where @xmath18 is the longitudinal sound velocity . we assume that at the `` transport '' stage , the temperature of the moving packet ( condensed@xmath19noncondensed particles ) is approximately equal to the lattice temperature , @xmath20 then we can consider the simplest case of @xmath21 and disregard the influence of all sorts of _ nonequilibrium _ phonons ( which appear at the stages of exciton formation , thermalization @xcite ) on the formed moving condensate . any theory of the exciton bec in cu@xmath0o has to point out some physical mechanism(s ) by means of which the key experimental facts can be explained . ( for example , the condensate moves without friction within a narrow interval of velocities localized near @xmath18 , and the shape of the stable macroscopic wave function of excitons resembles soliton profiles @xcite . ) here we explore a simple model of the ballistic exciton - phonon condensate . in this case , the general structure of the hamiltonian of the moving exciton packet and the lattice phonons is the following : @xmath22 here @xmath23 is the bose - field operator describing the excitons , @xmath24 is the field operator of lattice displacements , @xmath25 is the momentum density operator canonically conjugate to @xmath24 , and @xmath26 is the momentum operator . note that the hamiltonian ( [ ham1 ] ) is written in the reference frame moving with the exciton packet , i.e. @xmath27 and @xmath28 is the ballistic velocity of the packet . to derive the equations of motion of the field operators ( and generalize these equations to the case of @xmath29 ) , it is more convenient to start from the lagrangian . in the proposed model , the lagrangian density has the following form in the co - moving frame @xmath30 @xmath31 - _ g^ - ^2 ^ - _ 0 ( ^(x,t))^2 ( ( * x,t))^2 - _ 1(^ ( x,t))^3((x,t))^3 + @xmath32 + ( _ t)^2 -c_l^2 ( _ j_s)^2 -c_l^2_3 ( _ j_s)^3 -(_t_x + _ x _ t ) + v^2(_x)^2- @xmath31 @xmath33 where @xmath34 is the exciton `` bare''mass ( @xmath35 for @xmath36s excitons in cu@xmath0o ) , @xmath37 is the exciton - exciton interaction constant ( @xmath38 corresponds to the repulsive interaction between para - excitons in cu@xmath0o @xcite ) , @xmath39 is the crystal density , @xmath40 is the exciton - longitudinal phonon coupling constant , and @xmath41 . the energy of a free exciton is @xmath42 . although the validity of the condition @xmath43 ( @xmath44 is the exciton bohr radius ) makes it possible to disregard all the multiple - particle interactions with more than two participating particles in @xmath45 @xcite , we include the hard - core repulsion term originated from the three - particle interaction in @xmath46 , i.e. @xmath47 and @xmath48 ( for 3d case , one has to take @xmath49 because @xmath50 and @xmath51 ; see the discussion in @xcite . ) moreover , in the lagrangian of the displacement field , we include the first nonlinear term @xmath52 . ( the dimensionless parameter @xmath53 originates from taylor s expansion of an interparticle potential @xmath54 of the medium atoms . ) assuming that a dilute excitonic packet moves in a weekly nonlinear medium , we will not take into account more higher nonlinear terms in ( [ lagrangian ] ) . for simplicity s sake , we take all the interaction terms in @xmath55 in the _ local _ form and disregard the interaction between the excitons and transverse phonons of the crystal . note that the ballistic velocity @xmath10 is one of the parameters of the theory , and we will not take into account the excitonic normal component and velocity , i.e. @xmath56 , @xmath57 . this means that we choose the spatial part of the coherent phase of the packet , @xmath58 , to be in the simplest form , @xmath59 the equations of motion can be easily derived by the standard variational method from the following condition : @xmath60 indeed , after transforming the bose - fields @xmath61 and @xmath23 by @xmath62 we can write these equations as follows : @xmath63 @xmath64 @xmath65 we assume that the condensate of excitons _ exists_. this means that the following representation of the exciton bose - field holds : @xmath66 . here @xmath67 is the classical part of the field operator @xmath23 or , in other words , the condensate wave function , and @xmath68 is the fluctuational part of @xmath23 , which describes out - of - condensate particles . one of the important objects in the theory of bec is the correlation functions of bose - fields . the standard way to calculate them in this model ( the excitonic function @xmath69 , for example , ) can be based on the effective action or the effective hamiltonian approaches @xcite . indeed , one can , first , integrate over the phonon variables @xmath70 , get the expression for @xmath71 and , second , use @xmath72 ( or @xmath73 ) to derive the equations of motion for @xmath74 , @xmath68 , correlation functions , etc .. in this work we do not follow that way ; instead , we treat excitons and phonons equally @xcite,@xcite,@xcite . this means that the displacement field @xmath24 can have a nontrivial coherent part too , i.e. @xmath75 and @xmath76 , and the actual moving condensate can be an exciton - phonon one , i.e. @xmath77 . then the equation of motion for the classical parts of the fields @xmath23 and @xmath24 can be derived by use of the variational method from @xmath78 , in which all the fields can be considered as the classical ones . eventually we have @xmath79 @xmath80 @xmath81 notice that deriving these equations we disregarded the interaction between the classical ( condensate ) and the fluctuational ( noncondensate ) parts of the fields . that is certainly a good approximation for @xmath21 and @xmath82 cases @xcite . in this article , a steady - state of the condensate is the object of the main interest . in the co - moving frame of reference , the condensate steady - state is just the stationary solution of eqs . ( [ eq11]),([eq22 ] ) and it can be taken in the form @xmath83 where @xmath84 and @xmath85 are the real number functions , and @xmath86 is the coherent phase of the condensate wave function in the co - moving frame , see eq . ( [ superf ] ) . ( this phase can be taken equal zero if only a single condensate is the subject of interest . ) then , the following equations have to be solved , ( @xmath87 ) : @xmath88 @xmath89 note that in order to simplify eq . ( [ eq22 ] ) to eq . ( [ eq222 ] ) , we assumed only @xmath90 . in this model , it is enough to obtain localized solutions for the displacement field . solving eqs . ( [ eq111]),([eq222 ] ) in 3d space seems to be a difficult problem . however , these equations can be essentially simplified if we assume that the condensate is inhomogeneous along the @xmath91-axis only , that is @xmath92 note that the cross - section area of an excitation spot , @xmath93 , has to be basically constant across the sample cross - section . in this case , the problem can be considered as an effectively one - dimensional one . such an effective reduction of dimensionality transforms difficult ( nonlocal differential ) equations for the condensate wave function into a rather simple differential ones , and obtained in this way the effective 1d model for the condensate wave function @xmath94 conserves all the important properties of the `` parent '' 3d model . indeed , if @xmath95 , the following equations stand for the condensate ( @xmath96 ) : @xmath97 @xmath98 the last equation can be easily integrated , @xmath99 and solved relative to @xmath100 . here @xmath101 note that the medium nonlinearity parameter @xmath53 can be enhanced by the factor of the order of @xmath102 if the value of @xmath10 is less , but close to @xmath18 . ( for spatially localized solutions , @xmath103 and @xmath104 at @xmath105 , so that @xmath106 . ) if @xmath107 , we can always represent the solution of ( [ 11deq]),([11deqd ] ) in the following form @xmath108 ( indeed , the parameters of medium nonlinearity can be chosen as @xmath107 and @xmath109 @xcite . ) after substitution of ( [ dc ] ) into eq . ( [ 1deq ] ) , eqs . ( [ 1deq ] ) , ( [ 11deq ] ) can be rewritten as follows : @xmath110 @xmath111 where the interparticle interaction constants are renormalized as follows @xmath112 and more higher nonlinear terms are designated by @xmath113 . a small parameter in eq . ( [ kappa ] ) comes from the term @xmath114 where @xmath115 and @xmath116 is the mass of the crystal elementary cell . the effective two - particle interaction constant @xmath117 can be negative * if the velocity of the condensate lies inside the interval @xmath118 , where @xmath119 outside this interval , @xmath120 @xcite and the velocity @xmath121 can be called the first ` critical ' velocity in the model . the meaning of this velocity can be clarified by rewriting ( [ kappa ] ) in the dimensionless form , @xmath122 if @xmath123 , @xmath124 where @xmath125 and @xmath126 are the characteristic energy and the cross - section of two - particle collisions in the exciton subsystem . the following inequalities are true for excitons in a crystal @xmath127 and , usually , the value of the parameter @xmath128 is @xmath129 . for para - excitons in cu@xmath00 , however , we assume the ( effective ) value of @xmath130 can be estimated as @xmath131 , whereas the value of @xmath132 . this makes the inequality ( [ negative ] ) valid at , say , @xmath133 , or @xmath134 . thus , within the effective 1d model , the critical factor @xmath135 is the following ratio @xmath136 and , for the substances with @xmath137 , the regime with @xmath138 can be obtained at velocities _ reasonably close _ but not equal to @xmath18 , for example , beginning from some @xmath139 , ( @xmath140 ) . on the other hand , the effective three - particle interaction constant @xmath141 is always positive for crystals with @xmath142 . it can be represented in the dimensionless form as follows @xmath143 here we estimated the `` bare '' vertex of the three - particle collisions as @xmath144 and the same @xmath145 can be taken as a characteristic energy of collisions . the effective vertex @xmath146 is enhanced by the medium nonlinearity , and the both terms in the r.h.s . of ( [ three - part ] ) can be equally important at @xmath147 . note that in the case of _ strongly nonlinear lattices _ with excitons , the effective interaction vertices in ( [ 1deqmod ] ) ( @xmath148 , @xmath149 , etc . ) depend on the velocity @xmath10 and the parameters of medium nonlinearity ( @xmath53 , @xmath150 , etc . ) . then the effective exciton - exciton interaction can be strongly renormalized at sufficiently large gamma - factors @xmath151 and the vertices may change their signs as it can happen with @xmath117 . in this article , however , we consider the case of _ weakly nonlinear medium _ with excitons ( e.g. , a crystal with long living excitons ) . more accurately , this means that at velocities @xmath152 the effective vertex @xmath117 became @xmath153 , while the more higher vertices , such as @xmath154 and @xmath155 , do not change their sign ; they remain @xmath156 at @xmath157 . finally , to describe the weakly nonlinear case , it is enough to take into account the parameters @xmath158 and @xmath107 and neglect more higher nonlinearities ( @xmath113 and @xmath159 in ( [ 1deqmod ] ) , ( [ dcnl ] ) ) . in this study , we will consider the case of @xmath118 in detail . indeed , in the case of @xmath160 and @xmath161 , some localized solutions of eqs . ( [ 1deqmod]),([dcnl ] ) do exist . for example , the so - called ` bright soliton ' solution of ( [ 1deqmod ] ) exists if the generalized chemical potential is negative , @xmath162 , and @xmath163 . here @xmath164 for @xmath165 and @xmath166 , we can roughly estimate the effective vertex @xmath154 as @xmath167 then @xmath168 , and the more is the value of @xmath169 the less is the value of @xmath170 . the ` bright soliton ' solution of eq . ( [ 1deqmod ] ) can be represented in the following form @xmath171 here @xmath172 is some dimensionless parameter , and the generalized chemical potential @xmath173 is given by the formula @xmath174 like the chemical potential @xmath175 , the amplitude of the bright soliton , @xmath176 , satisfies @xmath177 and @xmath178 . for @xmath179 , the following approximation is valid @xmath180 and this formula can be used up to @xmath181 . then we can _ approximate _ the solution of ( [ 1deqmod ] ) by the following formulas @xmath182 @xmath183 the amplitudes of the exciton and phonon parts of the condensate , the characteristic width of the condensate , and the value of the effective chemical potential @xmath184 depend on the normalization of the exciton wave function @xmath185 . we normalize it in 3d space assuming that the characteristic width of the packet in the @xmath186-plane is sufficiently large , i.e. the cross - section area of the packet @xmath187 can be made equal to the cross - section area s of a laser beam and @xmath188 then we can write this condition as follows : @xmath189 where @xmath190 is the number of condensed excitons , and , generally , @xmath191 . applying this normalization condition , we get the following results @xmath192 here we used the following notations , @xmath193 , @xmath194 , where @xmath195 and @xmath196 . the formula ( [ amp1 ] ) is valid for @xmath197 . we assume that , at @xmath198 ( this is the important parameter ! ) , we always have @xmath199 then , the following inequalities are valid : @xmath200 and @xmath201 the characteristic length of the packet can be estimated from eq . ( [ genform ] ) as follows ( @xmath202 ) : @xmath203 @xmath204 therefore , at @xmath205 , we can roughly estimate @xmath206 and , for the average concentration of condensed excitons in the packet , @xmath207 , we have @xmath208 recall that the second part of the condensate , the displacement field @xmath209 , is of the same importance as the first part , the exciton wave function @xmath185 . the displacement field @xmath210 can be represented as follows @xmath211 to estimate its amplitude , @xmath212 , we have to estimate the parameter @xmath213 first . for @xmath214 and @xmath215 ( i.e. @xmath216 ) , we obtain @xmath217 if this parameter is small enough , such as @xmath218 , we can neglect the nonlinear corrections to the amplitude @xmath219 and to the shape of @xmath220 as well , @xmath221 thus , due to the validity of @xmath222 , there is almost no difference between the approximation @xmath223 where @xmath224 and the exact solution of the weakly nonlinear case with @xmath225 and @xmath107 . for @xmath226@xmath227 , @xmath228@xmath227 , we estimate @xmath229 . although the approximate solutions we used in this study are valid for @xmath230 , they can be used at @xmath231 for estimates . note that the effective chemical potential is a rather small parameter in this model , @xmath232 that is why the characteristic length , @xmath233 , see ( [ width ] ) , can be estimated as @xmath234 within the validity of approximations ( [ amp1 ] ) , ( [ mu1 ] ) . moreover , @xmath235 and the parameter @xmath172 in ( [ genform ] ) can be estimated as @xmath236 . in this case , one can neglect it in eq . ( [ solitonmod ] ) . returning to the laboratory reference frame , we can write the condensate wave function in the form ( see fig . 1 ) : @xmath237 @xmath238 where we count the exciton energy from the bottom of the crystal valence band , ( @xmath239 ) , and @xmath240 is the amplitude of the phonon part of condensate , @xmath241 to calculate the energy of the moving condensate within the lagrangian approach , ( see eq . ( [ lagrangian ] ) ) , we have to integrate the zeroth component of the energy - momentum tensor @xmath242 over the spatial coordinates . consequently , we have the following formula @xmath243 @xmath244 here we do not take into account a small correction to this energy due to the quantum depletion of the condensate ( @xmath245 and @xmath246 ) . then the result reads @xmath247 @xmath248 we will disregard the terms @xmath249 in @xmath250 , and the corrections @xmath251 in @xmath252 . then we can write @xmath253 and @xmath254 where @xmath255 note that the parameter @xmath256 is a rather small one , @xmath257 , so that the value of @xmath258 can be @xmath259 , and , roughly , @xmath260 . one can see that the exciton - phonon condensate carries a non - zeroth momentum , @xmath261 : @xmath262 @xmath263 @xmath264 thus , we obtain @xmath265 and estimate the parameter @xmath266 at @xmath216 , @xmath267 . = 400pt = 300pt to consider the stability of the exciton - phonon condensate moving in a lattice , one has to couple the excitons with different sources of perturbation , such as impurities , thermal lattice phonons , surfaces , etc .. in this work , however , we will not specify any source . instead , we consider the stability conditions in relation to creation ( emission ) of the condensate excitations that can be found in the framework of investigation of the low - energy excitations of the condensate itself . although the condensate wave function @xmath268 was obtained in the framework of the effective 1d model , we normalized it in 3d space . therefore , we can use this solution as a classical part in the following decomposition of 3d field operators in the co - moving frame : @xmath269 @xmath270 where @xmath271 . substituting the field operators of the form ( [ hatpsi]),([hatu ] ) into the lagrangian density ( [ lagrangian ] ) , we can write the later in the following form : @xmath272 where @xmath273 stands for the classical part of @xmath55 , and @xmath274 is the bilinear form in the @xmath275-operators . in the simplest ( bogoliubov ) approximation @xcite,@xcite , @xmath276 and , hence , the bilinear form @xmath274 defines the equations of motion for the fluctuating parts of the field operators . as a result , these equations are _ linear _ and can be written as follows : @xmath277 @xmath278 and @xmath279 @xmath280 @xmath281 @xmath282 @xmath283 the same approximation can be performed within the hamiltonian approach . indeed , decomposition of the field operators near their nontrivial classical parts leads to the decomposition of the hamiltonian ( [ ham1 ] ) itself , and as it was done with the lagrangian only the classical part of @xmath284 , @xmath285 , and the bilinear form in the fluctuating fields , @xmath286 , are left for examination : @xmath287 in the co - moving frame , @xmath288 , i.e. @xmath289 and the standard commutation relation , @xmath290 $ ] , has the form @xmath291= [ \ , \delta{\hat u}_{j}({\bf x},t),\,\,\rho\,\partial_{t}\delta{\hat u}_{s}({\bf x}',t ) - \rho v\,\partial_{x}\delta{\hat u}_{s}({\bf x}',t)\ , ] = i\hbar\,\delta({\bf x}-{\bf x'})\,\delta_{js}. \label{comrel1}\ ] ] however , the hamiltonian ( [ ham2 ] ) can be diagonalized and rewritten in the form : @xmath292 here , @xmath293 is the quantum correction to the energy of the condensate and the indexes @xmath294 and @xmath295 label the elementary excitations of the system . we assume the operators @xmath296 , @xmath297 are the bose ones . these operators describe two different branches of the excitations , @xmath298 , and they can be represented by the following linear combinations of the `` delta- operators '' : @xmath299 @xmath300 note that by analogy with the exciton - polariton modes in semiconductors@xcite,@xcite the excitations of the condensate ( [ movingcond ] ) can be considered as a mixture of exciton- and phonon - type modes . however , in this model , the phonons are fluctuations of the @xmath301-part of the condensate . the commutation relations between @xmath302-operators are bose ones , so that @xmath303= \delta_{ss'}\ ] ] lead to the following orthogonality condition @xmath304 since the @xmath302-operators ( see eq . ( [ ham3 ] ) ) evolve in time as simply as @xmath305 these operators ( and the frequencies @xmath306 ) are the eigenvectors ( and , correspondingly , the eigenvalues ) of the equations of motion ( [ deltapsi]),([deltau ] ) obtained within the lagrangian method . then , the time dependent `` @xmath275-operators '' in ( [ deltapsi]),([deltau ] ) can be represented by the following linear combinations of the @xmath302-operators : @xmath307 @xmath308 @xmath309 @xmath310 for @xmath311 , one has to change @xmath312 to @xmath313 in ( [ uvtransform2 ] ) . note that this _ ansatz _ is , in fact , a generalization of the u - v bogoliubov transformation . then we can rewrite eqs . ( [ alpha]),([alpha1 ] ) as follows ( @xmath298 ) @xmath314 @xmath315 and one of the orthogonality relations has the form ( @xmath316 ) @xmath317 @xmath318 the question we want to clarify is whether coupling between excitonic excitations and phonon excitations is important for understanding the condensate excitations . substituting ( [ uvtransform1])-([uvtransform2 ] ) into eqs . ( [ deltapsi]),([deltau ] ) , we obtain the following coupled eigenvalue equations @xcite : @xmath319 @xmath320 @xmath321 @xmath322{c}_{j , s}^{r}({\bf x})=0 , \,\,\,\,r=2,3 , \label{eigfunction3}\ ] ] @xmath323 @xmath324 { c}_{j , s}^{x}({\bf x})=0 . \label{eigfunction4}\ ] ] here we used the following notations @xmath325 @xmath326 to simplify investigation of the characteristic properties of the different solutions of eqs . ( [ eigfunction1])-([eigfunction3 ] ) , we subdivide the excitations ( [ uvtransform1])-([uvtransform2 ] ) into two major parts , the _ inside_-excitations and the _ outside_-ones . inside_-excitations are localized merely inside the packet area , i.e. @xmath327 and @xmath328 , whereas the _ outside_-excitations propagate merely in the outside area , i.e. @xmath329 and @xmath330 . for the outside collective excitations , the asymptotics of the low - lying energy spectrum can be found easily . indeed , if we assume that @xmath331 and @xmath332 in the outside packet area , the equations ( [ deltapsi ] ) and ( [ deltau ] ) are ( formally ) uncoupled . then , eq . ( [ deltapsi ] ) describes the excitonic branch of the outside - excitations with the following dispertion low in the co - moving frame : @xmath333 and @xmath334 in the laboratory frame of reference . then the exciton field operator , which describes the exciton condensate with _ one _ long - wavelength outside - excitation , has the following form : @xmath335 @xmath336 it is easy to see that such a collective excitation , @xmath337 can be interpreted as a free exciton with the energy and the ( quasi)momentum @xmath338 note that the condition @xmath339 can be violated at the velocities close to @xmath121 , whereas @xmath340 is always positive . then the question is whether @xmath341 really means the condensate instability in relation to the creation of outside excitations . for example , being unstable , the condensate could continuously emit outside - excitations , which form a sort of ` tail ' behind the localized packet . recall that the particle number @xmath342 is not conserved in quantum states with a condensate , and @xmath343 . however , for @xmath344 and @xmath345 , the following estimate is valid @xmath346 therefore , we can compare the condensate energy @xmath347 and the energy of the condensate that emits excitons , or , equivalently , the condensate with outside excitations , @xmath348 . for simplicity s sake , we consider @xmath349 different wave vectors , @xmath350 , to be close to each other , so that the values of @xmath351 and @xmath352 are well - defined . ( this is a model of how the instability tail could be formed . ) we obtain ( see eqs . ( [ energy])-([theta ] ) ) @xmath353 @xmath354 @xmath355 for the momentum of the moving condensate with the outside excitations , we have @xmath356 @xmath357 note that the energy and the momentum of the phonon part of the condensate change after exciton emission . we hypothesize that the transformation @xmath358 ( with the emission of outside excitons , see figs . 1,2 ) corresponds to the case in which the outside exciton and the outside acoustic phonon appear _ together_. indeed , in the @xmath359 limit ( i.e. @xmath360 ) , we approximately considered the condensate collective excitations as being uncoupled . however , the phonon @xmath361 can be emitted with the energy compensating the changement of @xmath362 in the phonon part of the condensate energy . moreover , the order of value of @xmath363 is typical for the low - energy acoustic phonons , @xmath364mev . if @xmath365 , the emitted phonon can compensate the changement of @xmath366 as well . the most interesting case is the backward emission of excitons , i.e. @xmath367 in the co - moving frame . then we can rewrite ( [ cond_and_exc1 ] ) as follows @xmath368 @xmath369 the moving condensate can be considered as a stable one in relation to emission of the outside excitations ( @xmath370 ) if such an emission gains energy , @xmath371 this means that the following inequality has to be valid @xmath372 this condition can be rewritten in the dimensionless form as follows : @xmath373 we argue that , even for velocities close to @xmath121 ( where @xmath374 can be @xmath375 , and the instability could appear as @xmath376 ) , inequality ( [ stabout ] ) seems to be always true in the long - wavelength approximation , @xmath377 . on fig . [ l - outside ] , the stable ballistic condensate is shown with its long - wavelength outside - excitations . note that the stability against large-@xmath378 modes can not be properly described within approximation ( [ gap1]),([outside ] ) . however , we can discuss this case within the inside - approximation . = 400pt = 300pt to simplify the calculation of _ inside_-excitation spectrum ( see eqs . ( [ eigfunction1])-([eigfunction3 ] ) ) we will use the semiclassical approximation @xcite . in this approximation , the excitations can be labeled by the wave vector @xmath12 in the co - moving frame , and the following representation holds : @xmath379 where the phase @xmath380 , and @xmath381 , @xmath382 , and @xmath383 are assumed to be smooth functions of @xmath384 in the inside condensate area . notice that the @xmath12- and @xmath384-representations are mixed here . this means that the operator nature of the fluctuating fields is actually dismissed within the semiclassical approximation . howevever , the orthogonality relations among @xmath385 , @xmath386 , and @xmath387 , @xmath388 , and , hence , among @xmath389 , @xmath390 , and @xmath391 , @xmath392 come from the bose commutation relations between the operators @xmath393 and @xmath394 @xcite,@xcite . for example , eq . ( [ uvtransform1 ] ) is modified as follows : @xmath395 and the _ inside_-excitation part of the elementary excitation term in eq . ( [ ham3 ] ) , @xmath396 , can be written as @xmath397 note that the semiclassical energy @xmath398 of the inside - excitation mode @xmath399 is supposed to be a smooth function of @xmath91 as well , ( at least , as smooth as @xmath400 , which is taken constant in the inside - approximation ) . although the low - lying excitations can not be properly described within the semiclassical approximation , we apply it here to calculate the low - energy asymptotics of the spectrum . in fact , within the approximation @xmath401 , all the important properties of such excitations can be understood by use of the semiclassical approach . there are two different types of the inside - excitations , the longitudinal excitations and the transverse ones . the later have the wave vectors @xmath12 perpendicular to the @xmath402-direction . for the sake of simplicity , we choose @xmath403 . then the vector @xmath404 has one nonzeroth component for such transverse excitations , @xmath405 . substituting _ ansatz _ ( [ semiclassical ] ) with @xmath406 and @xmath407 into eqs . ( [ eigfunction1])-([eigfunction3 ] ) , we transform these differential equations into the algebraic ones ( within the inside - approximation @xmath408 ) : @xmath409 @xmath410 @xmath411 c_{j , k}^{y}({\bf x})=0 . \label{eig}\ ] ] after some straightforward algebra , we can write out the equation that defines the spectrum of transverse exciton - phonon excitations in the inside - approximation : @xmath412 @xmath413=\ ] ] @xmath414\phi_{\rm o}^{2 } ( x)+ % \nu_{1}\phi_{\rm o}^{4}(x ) l(- k_{\perp}^{2})- \nu_{0}\phi_{\rm o}^{2 } ( { x } ) - 2\nu_{1}\phi_{\rm o}^{4 } ( { x } ) \,\right)\ , \,\frac{2\,\sigma_{0}^{2}}{\rho\,c_{l}^{2}}\,\phi_{\rm o}^{2}(x)\,(c_{l}^{2 } k_{\perp}^{2 } ) . \label{spectrum0}\ ] ] taking into account the momentum cut - off @xmath415 , which is defined as @xmath416 we can rewrite eq . ( [ spectrum0 ] ) as follows @xmath417 @xmath418 + f(x)+ \epsilon_{+}\right)\!\cdot\ ! \left ( \frac{\hbar^{2 } } { 2 m } [ k_{\perp}^{2}- { k}_{0}^{2 } ] + f(x)+2\nu_{0}\phi^{2}_{\rm o}(x ) + \epsilon_{+}'\right ) \,\right]=\ ] ] @xmath419 + f(x)+ \epsilon_{+}\right)\,2 \frac{\sigma_{0}}{m c_{l}^{2}}\,(\sigma_{0 } a_{l}^{3})\,\phi_{\rm o}^{2 } \,(c_{l}^{2 } k_{\perp}^{2 } ) , \,\,\,\,\,\ , \vert x \vert < l_{0}. \label{spectrum0tt}\ ] ] here @xmath420 , i.e. @xmath421 inside the condensate , and @xmath422 although eq . ( [ spectrum0tt ] ) can be solved exactly for the transverse excitation spectrum @xcite , taking into account the coupling term in the r.h.s . of ( [ spectrum0tt ] ) changes the values of excitation energies slightly , and the excitations can be approximately considered as of the pure excitonic ( @xmath423 ) or the pure phonon ( @xmath424 ) types . it is also useful to investigate _ asymptotics _ of the transverse inside excitations . for definiteness sake , we investigate the left side asymptotics of these excitations here , @xmath425 @xmath426 here @xmath427 , @xmath428 , @xmath429 , and @xmath430 are smooth functions of @xmath384 at @xmath431 . note that we introduced two components of @xmath432 to make eqs . ( [ eigfunction1])-([eigfunction4 ] ) self - consistent . let the equalities @xmath433 be valid . then the system of differential equations ( [ eigfunction1])-([eigfunction4 ] ) can be reduced to a system of algebraic ones , which are analogous to eqs . ( [ eigfunction01])-([eig ] ) . consequently , we can write out the equation for @xmath434 valid at @xmath431 , @xmath435^{2}\,- \,c_{l}^{2 } [ k^{2}_{\perp}-(2/l_{0})^{2}]\,\right ) \times\ ] ] @xmath436 + \tilde{f}(x ) \right)\cdot \left ( \frac{\hbar^{2 } } { 2 m } [ { k}^{2}_{\perp } - \tilde{k}_{0}^{2 } ] + \tilde{f}(x)+2\nu_{0}(2\phi_{\rm o}\exp(x / l_{0})\,)^{2}\right ) \,\right]= % \phi_{\rm o}^{2}(x)\right ) \,\right]= % ( 2\phi_{\rm o}\exp(x / l_{0})\,)^{2}\right ) \,\right]=\ ] ] @xmath437 + \tilde{f}(x)\,\right)\ , 2 \frac{\sigma_{0}}{m c_{l}^{2}}\,(\sigma_{0 } a_{l}^{3})\ , \bigl(2\phi_{\rm o}\exp(x / l_{0 } ) \,\bigr)^{2 } \,c_{l}^{2 } [ \,k^{2}_{\perp } - ( 2/l_{0})^{2 } ] , \label{spectrum1t}\ ] ] where @xmath438 , @xmath439 , and @xmath440 ( we neglected the terms , such as @xmath441 and @xmath442 in eqs . ( [ eigfunction1])-([eigfunction2 ] ) , and the terms @xmath443 in eq . ( [ eigfunction4 ] ) as well . ) obviously , the structure of eqs . ( [ spectrum0tt ] ) and ( [ spectrum1 t ] ) is the same . as the coupling between exciton and phonon branches is weak for the transverse inside - excitations ( see the r.h . sides of eqs . ( [ spectrum0tt ] ) and ( [ spectrum1 t ] ) ) and the effect of the finite width @xmath444 can be taken into account as the spatial dependence of the important parameters in @xmath445 , we use the following formula to estimate the low - energy excitation spectrum : @xmath446 @xmath447 here we take the inside - condensate - asymptotics of @xmath448 , @xmath400 , and @xmath449 to estimate @xmath450 . note that , for the inside - condensate excitations , the the low - energy limit means @xmath451 then , in the co - moving frame , the low - energy excitation spectrum @xmath452 may develop a gap of the order of @xmath175 , ( see eq . ( [ outside ] ) for comparison ) . thus , inside the condensate , we obtain a strong deviation of the collective excitation spectrum from both the simple excitonic one , @xmath453 , and the bogoliubov - landau spectrum @xmath454 . the case of the longitudinal excitations , @xmath455 , @xmath456 , is more difficult to analyze because the mode interaction is non - negligible in the low - energy limit . ( on fig . 1 . , a longitudinal inside - excitation is shown with the two possible directions of the wave vector @xmath457 . ) recall that the `` bare '' phonon modes , which can be written in the laboratory frame as @xmath458 with @xmath459 and @xmath460 , will be considered in the co - moving frame , @xmath461 . then , within the inside - condensate approximation , the following equation stands for the excitation spectrum : @xmath462 @xmath463 + f(x)+ \epsilon_{+}\right)\cdot \left ( \frac{\hbar^{2 } } { 2 m } [ k_{x}^{2}- { k}_{0}^{2 } ] + f(x)+2\nu_{0}\phi^{2}_{\rm o}(x ) + \epsilon_{+}'\right ) \,\right]=\ ] ] @xmath464 + f(x)+ \epsilon_{+}\right)\,2 \frac{\sigma_{0}}{m c_{l}^{2}}\,(\sigma_{0 } a_{l}^{3})\,\phi_{\rm o}^{2 } \,(c_{l}^{2 } k_{x}^{2 } ) , \,\,\,\,\,\ , \vert x \vert < l_{0}. \label{spectrum0ll}\ ] ] it is important to note that , unlike the case of transverse excitations , the values of @xmath465 and @xmath466 are of the same order of value at @xmath467 , and the inequality @xmath468 is valid in the low - energy limit . moreover , the two cases , @xmath469 ( @xmath19-case ) and @xmath470 ( @xmath471-case ) , are different as it can be seen from the l.h.s . of eq . ( [ spectrum0ll ] ) . in the low - energy limit , we can write the approximate solution of ( [ spectrum0ll ] ) as follows @xmath472 + f(x)+ \epsilon_{+}\right)\times\ ] ] @xmath473 + f(x)+ 2\nu_{0}\,\phi^{2}_{\rm o}(x ) \pm 2 \,q_{\pm } \gamma(v)\,\frac{\sigma_{0}}{m c_{l}^{2}}\,(\sigma_{0 } a_{l}^{3})\,\phi_{\rm o}^{2}(x ) + \epsilon_{+}'\right),\ ] ] where @xmath474 and @xmath475 . note that @xmath476 , whereas , for the phonon - type branch , @xmath477 . for @xmath478 , we have the following inequaluty for the excitonic branch , @xmath479 + f(x)+ \epsilon_{+}\right)\times\ ] ] @xmath480 + f(x)+ 2\,\widetilde{\nu_{0}}(v)\,\phi^{2}_{\rm o}(x ) + \epsilon_{+}'\right ) < \bigl(\hbar \omega^{(-)}_{{\rm ex } , \,k_{x } } \bigr)^{2 } < \bigl(\hbar \omega_{{\rm ex } , \,k_{x}}^{(0 ) } \bigr)^{2 } , \label{inequality1}\ ] ] where @xmath481 within the inside - approximation , and , for the phonon - type branch , we obtain @xmath482 . to derive the formulas for the amplitudes @xmath483 , @xmath484 , and @xmath485 of the excitonic branch , we use the following approximations @xmath486 and @xmath487 is modified as @xmath488 then we can rewrite the formulas for the excitonic excitation spectrum as follows @xmath489 recall that the orthogonality relation ( [ import22 ] ) can be used to normalize the amplitudes . within the inside - approximation , eq . ( [ import22 ] ) can be rewritten as follows ( @xmath490 ) @xmath491 and we have for the excitonic amplitudes @xmath492 @xmath493 here the effective condensate volume @xmath494 is used to normalize the u- and v - wave functions of the inside excitations , and @xmath495 . subsiquently , we get for @xmath496 the following approximate formulas @xmath497 to estimate the characteristic value of @xmath498 , we use @xmath499 and obtain @xmath500 the parameters @xmath501 characterize the relative weight of excitonic degrees of freedom in the considered branch of excitations . as the parameter @xmath502 at @xmath503 , the parameter @xmath504 can be estimated as @xmath505 . for @xmath470 , we obtain the following equation from ( [ ortuvc1 ] ) , @xmath506 within the stability area ( see the next subsection for an extended discussion ) , we estimate the ratio @xmath507 at @xmath508 . then , @xmath509 and @xmath510 . to go beyond the inside - approximation , the effect of inhomogeneous behavior of the longitudinal excitations can be considered . we use the following _ ansatz _ for the left side asymptotics ( see fig . 2 ) @xmath511 @xmath512 where @xmath513 and @xmath514 . then , like the case of transversal excitations ( see eq . ( [ spectrum1 t ] ) ) , we can write the equation for @xmath515 valid at @xmath431 , @xmath516 @xmath517 + \tilde{f}(x ) \right ) \left ( \frac{\hbar^{2 } } { 2 m } [ \bar { k}^{2}_{x } - { k}_{0}^{2 } ] + \tilde{f}(x)+2\nu_{0}\bigl(2\phi_{\rm o}\exp(x / l_{0})\,\bigr)^{2}\right ) \,\right]= % % \phi_{\rm o}^{2}(x)\right ) \,\right]= % % ( 2\phi_{\rm o}\exp(x / l_{0})\,)^{2}\right ) \,\right]=\ ] ] @xmath518 + \tilde{f}(x)\,\right)\ , 2 \frac{\sigma_{0}}{m c_{l}^{2}}\,(\sigma_{0 } a_{l}^{3})\ , \bigl(2\phi_{\rm o}\exp(x / l_{0 } ) \,\bigr)^{2 } \,c_{l}^{2}\tilde{k}^{2}_{x } , \label{spectrum1ll}\ ] ] where @xmath519 in the phonon parts of this equation and @xmath520 in the exciton parts of it ( @xmath521 ) . it is easy to see that eqs . ( [ spectrum0ll ] ) and ( [ spectrum1ll ] ) are in the continuity correspondence , i.e. they describe the same object . for example , the ( left side ) asymptotic behavior of @xmath522 can be obtained from the inside - condensate formulas by the substitute @xmath523 and @xmath524 as a result , we obtain @xmath525 that corresponds to the outside - excitation spectrum , eq . ( [ gap1 ] ) . the excitonic input into @xmath526 , the quantum depletion of the moving condensate , can be calculated by @xmath527 @xcite . to estimate this value , one can approximate @xmath528 as follows @xmath529 however , the summation @xmath530 implies @xmath531 within the semiclassical approximation . assuming that such an integration makes the difference among @xmath532 , @xmath533 and @xmath534 not essentially important , we use the following estimate for @xmath535 , @xmath536 then , the integration @xmath531 can be reduced to @xmath537 , and the main input ( @xmath538 ) can be estimated from the following formula @xmath539 using this density , we speculate that the _ localized _ depletion of condensate i.e. the number of particles that are out of the condensate but move with it coherently seems to be a small value . we obtain the following estimate ( @xmath540 and @xmath541 ) : @xmath542 where the factor before @xmath190 can be estimated as @xmath543 ( here we used eqs . ( [ need1]),([need2 ] ) within the approximation @xmath544 . ) note that there is no small @xmath12 input to the estimate of @xmath530 because , first , such excitations belong to the outside - excitation branch in our model , and , second , we use the approximation @xmath545 for them . to investigate the stability of the moving condensate in relation to the creation of inside - excitations , we calculate the energy of the condensate with the _ one _ inside excitation , @xmath546 , described by the following set : @xmath378 , @xmath547 , and @xmath548 , @xmath549 , @xmath550 . in this study , we analyze the stability inside the excitonic sector of our model . although the excitations were defined in the co - moving frame , calculations should be done in the laboratory frame . returning to the lab frame , we represent the exciton and phonon field functions as follows : @xmath551 where @xmath552 and @xmath553 see eq . ( [ outside ] ) for comparison . in this analysis , the inside excitations are not considered as fluctuations , and the ( average ) number of particles in the condensate and its energy are changed as @xmath554 and @xmath555 , respectively . however , these changes are not important if the number of excitation in a system is less than @xmath556 . they could be important in the case of instability of the moving condensate . the zeroth component of the energy - momentum tensor can be represented in the form @xmath557 where the first part corresponds to the condensate energy @xmath558 and the second part gives the energy of inside - excitations , @xmath559 . after substitution of ( [ urfin]),([urfin1 ] ) into @xmath560 , we have for the total energy @xmath561 @xmath562 @xmath563 where @xmath564 @xmath565 see eqs . ( [ cond_and_exc1]),([cond_and_exc1 ] ) for comparison . in this study , we discuss qualitatively the stability of the condensate in relation to the backward emission of inside - excitations , ( i.e. @xmath566 in the co - moving frame ) . to begin with , we consider the standard criterion , @xmath567 where @xmath568 corresponds to the low - lying excitations . the value of @xmath569 is taken within the inside - approximation , see eq . ( [ inequality1 ] ) , so that @xmath570 and @xmath571 . then , it is easy to conclude that the following inequality @xmath572 is valid in the low - energy limit if the effective chemical potential ( [ need2 ] ) is large enough , see eq . ( [ stable ] ) for comparison . more precisely , the ballistic velocity @xmath10 and the number of particles in the condensate , @xmath190 , have to be large enough , f.ex . , @xmath573 and @xmath574 , in order to the inequality @xmath575 can be satisfied . thus , for @xmath576 where @xmath577ev , one can expect conditions ( [ stable1]),([real_param ] ) to be valid . despite the condensate can be formed near @xmath578 in theory , f.ex . , with @xmath579 and @xmath544 , such a ballistic state seems to be unstable against the creation of inside - excitations . note that the critical ( landau ) velocity , @xmath580 , can be found as a solution of eq . ( [ cr_mu ] ) and @xmath581 . in fact , the parameter @xmath582 controls the stability / instability of the condensate , see eqs . ( [ stable]),([stable1 ] ) . analyzing ( [ stable ] ) , we did not take into account eq . ( [ stableornot ] ) . however , if the instability regime takes place , more than @xmath583 inside - excitations can appear . as the changes in @xmath584 because of @xmath585 are nonlocal ( in spite of creation of the localized excitations , see figs . 1 and 3 ) , a free acoustic phonon can appear in the system lattice@xmath19excitons _ together _ with appearance of the localized excitation @xmath586 . like the case of outside - excitations , we assume that @xmath587 , see eq . ( [ stableornot ] ) . then , only the term @xmath588 is important . in fact , this term leads to some renormalization of the values of the critical parameters , @xmath589 and @xmath580 . = 410pt = 300pt in this section , we address the problem of interaction between two moving condensates . this problem is essentially nonstationary , especially if the initial ballistic velocities of packets are different . within the quasi-1d conserving model , the following equations govern the dynamics of the two input packets ( we choose the reference frame moving with the slow packet ) : @xmath590 @xmath591 then , the initial conditions can be written in the explicit 1d form by using the exact solution of the model ( [ 1deq]),([11deq ] ) . note that the amplitudes of the stationary ballistic state , @xmath592 , were defined from the normalization condition and depend on the values of @xmath10 and @xmath190 . hence , the amplitudes of the `` input '' condensates for eqs . ( [ timedep1d]),([timedep1dd ] ) may not have the same values . = 400pt = 300pt in this study , we approach the problem of strong interaction between the condensates . therefore , we choose the nonsymmetric initial conditions , i.e. the amplitude and the velocity of the `` input '' packets are different , for example , @xmath593 and @xmath594 , see fig . [ l - inside11 ] . here , we reply on the experimental observation @xcite that , at @xmath595 , the ballistic velocity of the condensate depends on the power of a laser beam irradiating the crystal . if the exciton concentration in the first packet , @xmath596 , is close to the value of the bose condensation threshold , the exciton concentration in the second packet , @xmath597 , the velocity difference between condensates can reach @xmath598 . then , in the reference frame moving with the first ( slow ) packet , the initial conditions can be taken as the following : @xmath599 @xmath600 where @xmath601 , @xmath602 , @xmath603 , and @xmath604 is the ( initial ) time delay . as the second packet moves in this frame of reference , the regime of strong nonlinear interaction between the condensates is ( theoretically ) unavoidable . note that , even before collision , a time - dependent interference term in @xmath605 begins to influence the packet dynamics , see fig . 4 . for example , the r.h.s . of ( [ timedep1dd ] ) contains @xmath606 where @xmath607 and @xmath608 . the ratio @xmath609 can be of the order of @xmath610 , and the characteristic scale of fringes ( [ fringes ] ) is @xmath611 that is they are of the long - wavelength nature . to answer the question which model ( conserving ( [ timedep1d]),([timedep1dd ] ) or kinetic @xcite,@xcite one ) is more adequate to describe the packet collision , we have to compare the estimate of interaction time , @xmath612 and characteristic time scales of the processes @xmath613 driven by phonons or by @xmath614 interaction . note that some thermal phonons have to be excited in the system to assist such transitions , and the value of @xmath615 is of the order of scattering time of the exciton - la - phonon interaction ( although without any macroscopical occupancy ) @xcite . if processes ( [ process ] ) are driven by the lattice phonons , two phonons are necessary to satisfy the laws of conservation . for instance , we choose @xmath616 in ( [ process ] ) and obtain ( see eqs . ( [ cond_and_exc1]),([cond - and - exc1 ] ) ) @xmath617 @xmath618 @xmath619 although the second packet moves faster , @xmath620 , this state can be considered as a more stable ( and , thus , more preferable ) one for the excitons of the slow packet . indeed , the following inequality for the effective difference between the generalized chemical potentials seems to be valid @xmath621 and the absolute value of the l.h.s . of ( [ difference ] ) is @xmath622 . thus , within the quantum kinetic model , the relevant transition probabilities have to be calculated at least in the second order of perturbation theory , f.ex . , @xmath623 as a result , the system of two quantum boltzmann equations will describe the interaction process . unlike the boltzmann equations , eqs . ( [ timedep1d],[timedep1dd ] ) contain information about the quantum coherence between two condensates explicitly and , moreover , can describe the case of strong interaction . unlike the 1d nls equation that supports many - soliton solutions , eqs . ( [ timedep1d]),([timedep1dd ] ) are , in fact , quasi-1d ones , and @xmath38 in ( [ timedep1d ] ) . therefore , it is an open question what happens with two ( excitonic ) solitons after they collide in the crystal . in this work , we assume that the dominant process(es ) of condensate interactions is that one(s ) leading to @xmath624 . then , at the time scales @xmath625 , one solitonic packet can appear as a result of these processes . such a resultant ballistic packet can be approximately described by the steady - state one - soliton solution of eqs . ( [ eq111]),([eq222 ] ) with @xmath626 and the low of energy conservation , @xmath627 if @xmath628 , all the approximate solutions having been found in this study are valid to describe the resultant packet . as we prescribed the value of @xmath629 , we have to estimate the value of @xmath630 from eq . ( [ energy ] ) , ( generally , @xmath631 ) . moreover , we have to assume that the total momentum of the condensates , @xmath632 , may not be conserved because of lattice participation in such a condensate `` merger '' . however , the challenging question of the results of coherent packet collision needs further theoretical and experimental efforts . in this study , we considered a model within which the inhomogeneous excitonic _ condensate _ with a nonzero momentum can be investigated . the important physics we include in our model is the exciton - phonon interaction and the appearance of a coherent part of the crystal displacement field , which renormalizes the x - x interaction vertices . then , the condensate wave function and its energy can be calculated exactly in the simplest quasi-1d model , and the solution is a sort of davydov s soliton @xcite . we believe that the transport and other unusual properties of the coherent para - exciton packets in cu@xmath0o can be described in the framework of the proposed model properly generalized to meet more realistic conditions . we showed that there are two critical velocities in the theory , namely , @xmath121 and @xmath580 . the first one , @xmath121 , comes from the renormalization of two particle exciton - exciton interaction due to phonons , and the bright soliton state can be formed if @xmath595 . then , the important parameter , which controls the shape and the characteristic width of the condensate wave function , is @xmath633 , see eqs . ( [ mu1]),([need2 ] ) . the second velocity , @xmath580 , comes from use of landau arguments @xcite for investigation of the dynamic stability / instability of the moving condensate . in fact , the important parameter , which controls the emission of excitations , is @xmath634 , see eq . ( [ cr_mu ] ) . then , within the semiclassical approximation for the condensate excitations , we found more close @xmath10 is to @xmath18 ( @xmath635 ) more stable the coherent packet is . it is interesting to discuss the possibility of observation of an instability when the condensate can be formed in the inhomogeneous state with @xmath636 , but with @xmath637 , or , better , @xmath638 . such a coherent packet has to disappear during its move through a single pure crystal used for experiments . as the shape of the moving packet depends on time , the form of the registered signal may depend on the crystal length changing from the solitonic to the standard diffusion density profile . we found that the excited states of the moving exciton - phonon condensate can be described by use of the language of elementary excitations . although the possibility of their direct observation is an unclear question itself ( and , thus , the question about the gap in the excitation spectrum is still an open one ) , the stability conditions of the moving condensate can be derived from the low - energy asymptotics of the excitation spectra at @xmath345 . however , the stability problem is not without difficulties @xcite,@xcite . one can easily imagine the situation when the condensate moves in a very high quality crystal , but with some impurity region prepared , f.ex . , in the middle of the sample . in this case , the excitonic superfluidity can be examined by impurity scattering of the ballistic condensate . indeed , such impurities could bound the noncondensed excitons , which always accompany the condensate , and could mediate , for instance , the emission of the outside excitations . the last process may lead to depletion of the condensate and , perhaps , some other observable effects , such as damping , bound exciton pl , etc .. one of the authors ( d.r . ) thanks i. loutsenko for helpful discussions , a. mysyrowicz for useful comments , and e. benson and e. fortin for providing the results of their work before publishing . a. l. ivanov , h. haug , and l. v. keldysh , phys . reports * 296 * , 237 ( 1998 ) ; h. haug , a. l. ivanov , and l. v. keldysh , _ nonlinear optical phenomena in semiconductors and semiconductor microstructures _ , ( world scientific , 1999 ) .	we explore a nonlinear field model to describe the interplay between the ability of excitons to be bose - condensed and their interaction with other modes of a crystal . we apply our consideration to the long - living para - excitons in cu@xmath0o . taking into account the exciton - phonon interaction and introducing a coherent phonon part of the moving condensate , we derive the dynamic equations for the exciton - phonon condensate . these equations can support localized solutions , and we discuss the conditions for the moving inhomogeneous condensate to appear in the crystal . we calculate the condensate wave function and energy , and a collective excitation spectrum in the semiclassical approximation ; the inside - excitations were found to follow the asymptotic behavior of the macroscopic wave function exactly . the stability conditions of the moving condensate are analyzed by use of landau arguments , and landau critical parameters appear in the theory . finally , we apply our model to describe the recently observed interference and strong nonlinear interaction between two coherent exciton - phonon packets in cu@xmath0o . -10 mm -0.5 cm * bosons in a lattice : * * exciton - phonon condensate in cu@xmath0o * d. roubtsov and y. lpine _ groupe de recherche en physique et technologie des couches minces , _ _ dpartement de physique , universit de montral , _ c.p . 6128 , succ . centre - ville , montreal , pq , h3c3j7 , canada e - mail : [email protected] pacs numbers : 71.35.+z , 71.35.lk
statistical physics is the branch of theoretical physics which provides a microscopic basis for the macroscopic theory of thermodynamics . equilibrium statistical physics makes use of statistical ensembles , like the microcanonical or the canonical one . the basic thermodynamic function in the microcanonical ensemble is the entropy @xmath6 as a function of some energy @xmath2 , magnetization @xmath3 , and/or further arguments , whereas in the canonical ensemble this role is played by the canonical free energy @xmath7 as a function of the temperature @xmath8 , a magnetic field @xmath9 , and/or further arguments . the computation of thermodynamic functions from statistical mechanics is a formidable , in most cases impossible task in general , but , as a rule of thumb , it is more difficult in the microcanonical ensemble . this is the reason why , although the microcanonical ensemble is the most fundamental one of the statistical ensembles , one often resorts to the calculation of the canonical free energy @xmath10 instead . this is a reasonable way to proceed , as long as both options yield equivalent results . if this is not the case , one speaks of ensemble nonequivalence , and the occurrence of non - concave microcanonical entropy functions in systems with long - range interactions is a hallmark of this situation . then , unlike in standard thermodynamics where the concavity of the entropy is assumed , the entropy @xmath11 can not be obtained from the free energy @xmath10 by means of a legendre transform . only in recent years the community of statistical physicists became aware that large deviation theory , a branch of probability theory , is a powerful tool to compute , among others , microcanonical quantities , at least for some classes of systems ( see @xcite for a mathematical treatise on large deviation theory with applications to statistical mechanics , or @xcite for a less formal presentation of the subject ) . explicit calculations of microcanonical quantities by means of large deviation theory can be found in @xcite for some models of physical interest ( the infinite range potts model , the hamiltonian mean - field model , and the colson - bonifacio model of a free electron laser ) . we start by briefly reviewing some results of large deviation theory and their relation to statistical mechanics in sec . [ sec : ldt ] . in sec . [ sec : model ] , the microcanonical entropy of the @xmath1-model without interaction is calculated using large deviation theory . from this result , the entropy of the mean - field @xmath1-model in the presence of interactions is deduced . the physical implications of the result are discussed , followed by some conclusions in sec . [ sec : concl ] . this section contains a brief , informal review of some statements of large deviation theory and how they can be applied to statistical mechanics . consider a sequence of independent , identically distributed ( i.i.d . ) random variables @xmath12 with probability distribution @xmath13 , the empirical mean @xmath14 , and the mean ( or expectation ) value @xmath15:=\int\,{\mathrm{d}}x\,x\,p(x)$ ] . the law of large numbers tells us that , in the limit of large @xmath16 , the probability @xmath17 of an event with @xmath18 converges to zero . the form of this convergence is the central issue of large deviation theory , thus allowing predictions about very unlikely events . under some prerequisites on the random variables @xmath19 , cramr @xcite found that , in the limit of large @xmath16 , the convergence is of exponentially decaying form in @xmath16 , @xmath20).\ ] ] the so - called rate function @xmath21 depends on the value @xmath22 the empirical mean has to meet . comparing to the definition of the boltzmann entropy , @xmath23,\ ] ] we see that @xmath24 is identical to @xmath25 up to a physically irrelevant additive constant . under suitable conditions on @xmath19 , large deviation theory gives an instruction how to derive the rate function ( and therefore the microcanonical entropy ) : @xmath26,\ ] ] where @xmath27 denotes the usual scalar product and @xmath28 is obtained by solving the conditional equation @xmath29 = { \mathbbm{e}}\left[x\,{\mathrm{e}}^{(t_x , x)}\right].\ ] ] the mean - field @xmath1-model we want to study is characterized by a potential energy function of the form @xmath31 where @xmath16 is the number of particles , @xmath5 is a coupling constant , and @xmath32 denotes the position in configuration space . our aim is to compute the microcanonical configurational entropy of this model . first , we look at the @xmath1-model without interaction , where @xmath16 particles move independently in an on - site potential of the form @xmath33 . the potential energy is @xmath34 the first step in calculating the configurational entropy is to translate our problem into the language of the previous section . considering the coordinates @xmath35 as random variables with flat distribution on some finite interval , @xmath36$},\\ 0 & \text{else } , \end{cases}\ ] ] other random variables @xmath37 with potential energy and displacement of the @xmath38-th particle as components can be constructed from the @xmath39 . this form of is motivated by the fact that it allows to express the potential energy function by the empirical mean @xmath40 , where @xmath41 the associated macroscopical variables @xmath42 are the mean potential energy per particle @xmath43 and the mean magnetization ( or displacement ) per particle @xmath3 . then , following eqs . and , the entropy of the non - interacting @xmath1-model can be written as @xmath44 where @xmath45 and @xmath46 are obtained by solving the conditional equations @xmath47 containing the integrals @xmath48 we let @xmath49 tend to infinity in the following . solving for @xmath50 and @xmath51 seems not feasible analytically . however , besides a numerical solution ( see fig . [ figuresmz ] ) , some general properties of @xmath52 may be of interest . from large deviation theory we know that @xmath52 is a concave , infinitely many times differentiable function . its domain can be determined by the following argument : the energy @xmath53 of any single particle is not bounded above , but it is bounded below by @xmath54 . therefore also the mean potential energy @xmath43 is restricted to the interval @xmath55 . for any value of @xmath56 , there exists a maximal and a minimal value of the magnetization @xmath57 which is attained when @xmath58 for all @xmath59 . so @xmath60 is determined by @xmath61 . starting from a microstate with @xmath62 , we can always find another microstate with @xmath63 $ ] by changing the sign of an arbitrary fraction of the values of the @xmath64 . for an illustration of the domain of @xmath52 see fig . [ figuresmz ] . for any fixed value of @xmath65 , the maximum of @xmath66 with respect to @xmath3 is located at zero magnetization . turning to the study of the interacting model , we can now profit from our particular choice of the random variables @xmath19 in eq . , resulting in the macroscopic variables @xmath67 . from the form of the potential energy function @xmath68 , the mean potential energy per particle can be written as @xmath69 and for the corresponding macroscopic variable @xmath2 the equality @xmath70 holds . as a consequence , the entropy @xmath0 of the mean - field @xmath1-model is obtained from the entropy of the non - interacting model by a simple transformation of variables , @xmath71 a plot of the resulting function is shown in fig . [ figuresmv ] . for large enough fixed values of the potential energy @xmath2 , the maximum of @xmath11 with respect to @xmath3 is again located at zero magnetization . below a critical value @xmath72 of the potential energy , however , this is not the case anymore , indicating the occurrence of a phase transition . the transition is found to be a continuous one , and @xmath72 is determined by a change of curvature of @xmath11 with respect to @xmath3 along the line of zero magnetization , @xmath73 from the results of sec . [ sec : no_int ] , the critical potential energy @xmath72 in dependence of the coupling constant @xmath5 can be written as @xmath74 where @xmath75 is defined implicitly by @xmath76 it follows immediately that @xmath77 and , since @xmath78 , that @xmath79 for all @xmath80 . the integral @xmath81 in can be rewritten in terms of modified bessel functions of the first kind @xmath82 , yielding @xmath83 this allows to derive asymptotic expansions of @xmath4 for small and large positive @xmath5 , respectively , from the known expansions of these bessel functions . [ [ small - j0 ] ] small @xmath80 : + + + + + + + + + + + + + + + + + + + + + + + + inserting the asymptotic expansion of for large negative @xmath75 into , solving for @xmath84 and substituting into finally yields @xmath85 [ [ large - j0 ] ] large @xmath80 : + + + + + + + + + + + + + + + + + + + + + + + + analogously , an expansion of for small negative @xmath75 yields an expression for the critical energy in the limit of large @xmath5 , @xmath86 with @xmath87 . together with the numerical result from the evaluation of eqs . and , these two asymptotic expansions are plotted in fig . [ figurecritical ] . the microcanonical entropy @xmath0 as a function of the potential energy @xmath2 and the magnetization @xmath3 of the mean - field @xmath88-model has been calculated using a large deviation technique . for such long - range interactions , and in contrast to standard thermodynamics , the entropy need not be a concave function . in fact , for a ferromagnetic coupling @xmath80 , a non - concavity in @xmath11 is found , and therefore , when comparing this result to the canonical free energy @xmath89 as a function of the temperature @xmath8 and the magnetic field @xmath9 , nonequivalence of ensembles is observed . for any fixed value of @xmath2 above a critical energy @xmath72 , the maximum of @xmath0 with respect to @xmath3 is attained at @xmath90 , whereas below @xmath72 it is attained at @xmath91 , corresponding to a continuous order phase transition at @xmath72 . an exact , implicit expression for the critical energy @xmath4 as well as expansions for small and large positive @xmath5 are derived . maximizing @xmath0 over @xmath3 gives a concave entropy function @xmath92,\ ] ] showing ensemble equivalence when compared to the canonical free energy @xmath93 $ ] . in fact , the critical energy @xmath4 from our microcanonical calculation corresponds to the critical temperature @xmath94 of the canonical result reported in @xcite .	a large deviation technique is used to calculate the microcanonical entropy function @xmath0 of the mean - field @xmath1-model as a function of the potential energy @xmath2 and the magnetization @xmath3 . as in the canonical ensemble , a continuous phase transition is found . an analytical expression is obtained for the critical energy @xmath4 as a function of the coupling parameter @xmath5 . phase transition , mean - field @xmath1-model , entropy , large deviation , ensemble equivalence
it is a classical result that the existence of certain algebraic structures on a space can determine whether or not it is an @xmath3-fold loop space ( where here @xmath3 is possibly @xmath4 ) . for instance , beck @xcite showed that a space is an @xmath3-fold loop space if and only if it is an algebra over the monad @xmath5 . in @xcite may found a much simpler description using monads which come from operads . such monads have the useful property that they are _ finitary _ : completely determined by their restriction to finite sets . finitary monads and their algebras correspond to ( lawvere ) theories and their algebras ( * ? ? ? * ii.4 ) . [ detectable ] a * theory * @xmath6 is a based simplicial category with objects @xmath7 such that @xmath8 is the @xmath9-fold product of @xmath10 . in particular , @xmath11 is the basepoint . an * algebra * over @xmath6 is a based , product preserving functor @xmath2 from @xmath6 into based simplicial sets . the * underlying space * of @xmath2 is @xmath12 . so an algebra over a theory consists of a space @xmath12 along with operations @xmath13 that satisfy certain relations which are parameterized by the theory @xmath6 . the above discussion is saying that loop spaces are detected by theories , in the following sense . a space @xmath0 is * detectable * if there is a theory @xmath6 with the property that a space @xmath2 is weakly equivalent to the underlying space of a @xmath6-algebra if and only if @xmath2 is weakly equivalent to @xmath14 for some @xmath15 . one of the upshots of a space being detectable is that algebras over a theory are closed under various operations . for example , let @xmath16 be a functor from spaces to spaces which preserves weak equivalences and preserves products up to weak equivalences . then ( * ? ? * cor 1.4 ) says that if @xmath2 is weakly equivalent to an algebra over a theory @xmath6 , so is @xmath17 . so if @xmath0 is detectable , applying such a functor @xmath16 to a mapping space of the form @xmath18 gives a space which is weakly equivalent to @xmath19 for some space @xmath20 . in particular , if @xmath0 is detectable , then the localization @xmath21 with respect to any map @xmath22 is weakly equivalent to @xmath19 for some space @xmath20 . in @xcite , it is shown that if @xmath0 is a finite , pointed cw - complex with the property that its mapping spaces are closed under localization in this way , then @xmath0 has the rational homotopy type of a wedge of spheres which are all the same dimension . based on this result , it is not clear if _ any _ spaces are detectable other than wedges of @xmath3-spheres . on the contrary , we show that any space which satisfies a certain smallness condition is detectable . in particular , this includes spheres localized at a set of primes . our main result is the following : [ maintheorem ] let @xmath23 be the @xmath24-local sphere for @xmath24 a set of primes , and @xmath25 . there is a theory @xmath26 with the property that a space @xmath2 is weakly equivalent to @xmath27 for some @xmath15 if and only if @xmath2 is weakly equivalent to the underlying space of an @xmath26-algebra . as in @xcite , we will actually prove the stronger statement that there is a quillen equivalence between the category of algebras over the theory @xmath26 and the right bousfield localization of spaces with respect to @xmath23 . + _ notation _ * we work in the category of pointed simplicial sets , which we denote by @xmath28 . * for a set of primes @xmath24 and @xmath25 , the @xmath24-local sphere @xmath23 is the singularization of the mapping telescope : + @xmath29 + where @xmath30 are the maps whose degree are the positive integers relatively prime to the primes in @xmath24 . * we freely use the language of model categories . for an introduction see @xcite . i wish to thank my advisor , bernard badzioch , for sparking my interest in homotopy theory and algebraic structures , and for his patience and comments through several versions of this paper . i also would like to thank bruce corrigan - salter and alyson bittner for many illuminating conversations . the category of pointed simplicial sets is equipped with a pointed mapping space given by : @xmath31 where the simplicial structure is encoded by the cosimplicial structure of @xmath32 , which is the pointed cosimplicial space sending each @xmath33 \in \delta$ ] to the standard @xmath3-simplex with a disjoint basepoint . for any pointed simplicial set @xmath0 , we define a theory @xmath1 by : @xmath34 that this actually is a theory is probably easiest seen by the fact that the adjunction between pointed simplicial sets and topological spaces induces isomorphisms of simplicial mapping spaces : @xmath35 thus we may view our theory as the opposite of the full subcategory of the ( pointed simplicial ) category of pointed topological spaces consisting of wedges of copies of @xmath36 . since @xmath37 is the @xmath3-fold coproduct of @xmath36 , it becomes the @xmath3-fold product in the opposite category . [ canonical ] let @xmath0 be a space . the theory @xmath1 just described is called the * canonical theory associated to * @xmath0 . in @xcite , it was shown that for @xmath38 , the canonical theory detects spheres in the sense of definition [ detectable ] . indeed , for any space @xmath0 , the mapping space @xmath18 is the underlying space of the @xmath1 algebra : @xmath39 @xmath40 the content of @xcite was that , in the case of a sphere , every algebra over the canonical theory @xmath41 has underlying space weakly equivalent to an @xmath3-fold loop space . we recall the homotopy theory of algebras over a theory @xmath6 . let @xmath42 denote the category of algebras over a theory @xmath6 , where the morphisms are natural transformations . then taking the underlying space gives a functor @xmath43 from the category of algebras over @xmath6 to pointed simplicial sets . it has a left adjoint @xmath16 , the free algebra functor , and this adjunction is used to lift the quillen model category of pointed simplicial sets to the category of algebras . @xcite[algmod ] there is a model category structure on @xmath44 , where a map @xmath45 between algebras is a : 1 . weak equivalence if @xmath46 is a weak equivalence in @xmath28 2 . fibration if @xmath46 is a fibration in @xmath28 3 . cofibration if @xmath47 has the left lifting property with respect to the acyclic fibrations . moreover , with this model structure , the free / forgetful adjunction @xmath48 is a quillen adjunction . the fibrant algebras are the algebras whose underlying spaces are fibrant simplicial sets . in the remainder of this section , we define a useful cofibrant replacement of an algebra . let @xmath2 be an algebra over a theory @xmath6 . define the simplicial @xmath6-algebra @xmath49 to be @xmath50 in simplicial degree @xmath3 . the face and degeneracy maps are defined using the unit @xmath51 and counit @xmath52 of the adjunction . specifically , we have @xmath53 @xmath54 the counit @xmath55 induces a map @xmath56 the map @xmath57 is called * bar resolution of * @xmath2 . [ res ] the bar resolution @xmath57 is a cofibrant replacement in the category of algebras . the fact that @xmath58 is a weak equivalence follows from a standard extra degeneracy argument , as in ( * ? ? ? * proposition 9.8 ) . to complete the proof , we need to show that @xmath49 is cofibrant . we will do this by showing that @xmath49 is a reedy cofibrant algebra in the category of simplicial algebras . then geometric realization , being a left quillen adjoint , sends reedy cofibrant simplicial algebras to cofibrant algebras . let @xmath59 be the full subcategory of @xmath60 consisting of the degeneracy maps and only positive face maps . like @xmath60 , this is a reedy category . consider the functor @xmath61 from @xmath59 to @xmath62 which is @xmath63 in simplicial degree @xmath3 and whose face maps and degeneracy maps are similar to @xmath49 . all of the latching maps of @xmath64 are cofibrations , and applying @xmath16 we see that the restriction of @xmath49 to @xmath65 is reedy cofibrant . by ( * proposition 3.17 ) , the simplicial algebra @xmath49 is reedy cofibrant as well . the mapping space algebra @xmath66 in the previous section defines a functor from @xmath28 to the category of algebras @xmath44 whose morphisms are natural transformations . this functor has a left adjoint , which we call @xmath67 , and this adjunction passes to the level of homotopy categories . let @xmath0 be a space , and let @xmath1 be the canonical theory from definition [ canonical ] . then the adjunction @xmath68{b^a } \mathbf{alg}^{t_a}$ ] is quillen . it suffices to show that the right adjoint preserves fibrations between fibrant objects and acyclic fibrations . if @xmath69 is a fibration between fibrant objects , or an acylic fibration , then so is the induced map @xmath70 in other words @xmath71 is a fibration or acyclic fibration , and by definition of the model structure on algebras , so is @xmath72 . we will see that for spaces @xmath0 which satisfy a smallness condition ( see definition [ homotopyprojective ] ) we can define a model category structure on @xmath28 so that this adjunction becomes a quillen equivalence . when this is possible , the right adjoint @xmath66 has to reflect weak equivalences between fibrant objects . the following model category structure , called the right bousfield localization of @xmath28 with respect to @xmath0 , can be thought of as the most efficient model category which makes this happen . we will denote it @xmath73 ( * ? ? ? * theorem 5.1.1 ) let @xmath0 be a pointed simplicial set . there is a model category structure on @xmath28 , where a map @xmath69 is a : 1 . weak equivalence if the induced map latexmath:[$\mathbf{map_}(a , weak equivalence . 2 . fibration if it is a kan fibration . cofibration if it has the left lifting property with respect to all acyclic fibrations . this follows from a general theorem on existence of right bousfield localizations ( ? ? ? * theorem 5.1.1 ) , since the model category of pointed simplicial sets is proper and cellular ( * ? ? ? * proposition 5.1.8 ) by @xcite , the cofibrant spaces in @xmath73 are the @xmath0-cellular spaces , which we define now . a nonempty class of simplicial sets is said to be * closed * if it is closed under homotopy colimits and weak equivalences . the smallest closed class which contains a simplicial set @xmath0 is called the class of * @xmath0-cellular spaces * , denoted * cell * @xmath0 . the @xmath0-cellular spaces are roughly the spaces built from @xmath0 . for example , the @xmath75-cellular spaces are the @xmath76-connected simplicial sets . for us , the most important example is the following . let @xmath25 , and @xmath24 be a set of primes . then @xmath77 is the class of @xmath76-connected @xmath24-local cell - complexes . by definition , @xmath77 is the smallest class of simplicial sets containing @xmath23 which is closed under weak equivalences and homotopy colimits . the constructions in @xcite show that it is possible to build any @xmath24-local space by homotopy colimits . it is shown in @xcite that it is enough to consider filtered colimits and homotopy pushouts . since the property of being @xmath76-connected and @xmath24-local is determined by reduced homology , any homotopy pushout or filtered colimit of @xmath76-connected @xmath24-local spaces is still @xmath76-connected and @xmath24-local . the final ingredient in our recognition principle is that @xmath0 needs to satisfy a smallness condition . [ homotopyprojective ] let @xmath78 be a closed class . a pointed simplicial set @xmath0 is called * homotopy projective relative to * @xmath78 if @xmath79 commutes with homotopy colimits of filtered or simplicial diagrams which take their value in @xmath78 . a space @xmath0 is called * homotopy self - projective * if it is homotopy projective relative to * cell * @xmath0 . the spheres @xmath75 are homotopy self - projective . @xmath80 commutes with homotopy filtered colimits because @xmath75 is a finite simplicial set . the class * cell * @xmath75 is exactly the class of @xmath76-connected spaces , so it follows from the bousfield - friedlander theorem ( * ? ? ? * theorem b.4 ) that @xmath80 commutes with homotopy colimits of @xmath76-connected simplicial spaces . [ hoproj ] for @xmath25 and @xmath24 a set of primes , the @xmath24-local sphere @xmath23 is homotopy self - projective . the map : @xmath81 induced from the localization map @xmath82 is a weak equivalence for @xmath24-local spaces @xmath83 . let @xmath84 be either a filtered diagram or else a simplicial diagram which takes its values in * cell * @xmath23 . we have the following commutative diagram : ( m)[matrix of math nodes , row sep=2em , column sep=4em , minimum width=2em , text height = 1.5 ex , text depth = 0.25 ] ( s^n_p , ) & ( s^n_p , ) + ( s^n , ) & ( s^n , ) + ; ( m-1 - 1 ) edge node [ above ] @xmath85 ( m-1 - 2 ) ( m-2 - 1 ) edge node [ below ] @xmath86 ( m-2 - 2 ) ( m-1 - 1 ) edge node [ left ] @xmath87 ( m-2 - 1 ) ( m-1 - 2 ) edge node [ right ] @xmath87 ( m-2 - 2 ) ; the left and right maps are weak equivalences because @xmath88 takes its value in @xmath24-local spaces . the bottom map is a weak equivalence because @xmath75 is homotopy self - projective and @xmath88 takes its values in @xmath76-connected spaces . hence the top map is also a weak equivalence . these are actually the only homotopy self - projective spaces we know of , and we think it is likely that wedges of @xmath3-spheres are the only finite spaces which are homotopy self - projective . on the other hand , we can say that the property of being homotopy self - projective is stable . if @xmath0 is connected and homotopy self - projective , so is @xmath89 . a space @xmath2 is in @xmath90 if and only if @xmath91 is in @xmath92 @xcite . so if @xmath93 is some filtered or simplicial diagram which takes its values in @xmath90 , then the diagram @xmath94 obtained by postcomposition takes its values in @xmath92 . since @xmath0 is homotopy self - projective @xmath95 is a weak equivalence . since @xmath88 takes its values in connected spaces , and since @xmath96 is homotopy self - projective there is a weak equivalence @xmath97 in this section we prove [ maintheorem ] . it will follow easily from the following is a _ right _ quillen adjoint , so that this is not automatic . however @xmath73 is a simplicial model category with the same simplicial structure as @xmath28 ( * ? ? ? * theorem 5.1.2 ) . so @xmath66 preserves fibrations between fibrant objects and acyclic fibrations . next , since @xmath66 reflects weak equivalences between fibrant objects , it is enough to show that the unit map @xmath100 is a weak equivalence for all cofibrant @xmath2 . we split the proof into cases : * 1 . * consider first the case when @xmath101 , where @xmath102 is the set based at @xmath103 . for any space @xmath83 , the composition @xmath104 is the mapping space @xmath105 , so that in the other direction @xmath106 is @xmath107 . thus , applying the forgetful functor to the map @xmath108 is isomorphism : ( m)[matrix of math nodes , row sep=4em , column sep=5em , minimum width=4em , text height = 2.5 ex , text depth = 1 ] _ ^ fu_(x ) & ^a \|b^a _ ^fu_(x)\| + x & ^a \|b^a x\| + ; ( m-1 - 1 ) edge node [ above ] @xmath121 ( m-1 - 2 ) ( m-2 - 1 ) edge node [ below ] @xmath122 ( m-2 - 2 ) ( m-1 - 1 ) edge node [ left ] @xmath58 ( m-2 - 1 ) ( m-1 - 2 ) edge node [ right ] @xmath123 ( m-2 - 2 ) ; the bar resolution has the form from step 3 , so that @xmath121 is a weak equivalence . by theorem [ res ] , the map @xmath58 is a weak equivalence . finally , @xmath66 preserves weak equivalences between fibrant objects , so that @xmath122 is a weak equivalence as well . we showed in section 2 that if a space @xmath2 is weakly equivalent to @xmath18 for some @xmath15 , then @xmath2 is weakly equivalent to the underlying space of the algebra @xmath124 . conversely , suppose a space @xmath2 is weakly equivalent to the underlying space of an algebra @xmath125 , then @xmath2 is also weakly equivalent to the underlying space of a cofibrant replacement @xmath126 of @xmath125 . by [ q ] , there is a weak equivalence of algebras @xmath127 . by the definition of weak equivalences between algebras , it follows that @xmath2 is weakly equivalent to @xmath128 .	we show that for spaces @xmath0 that satisfy a certain smallness condition , there is a lawvere theory @xmath1 so that a space @xmath2 has the structure of a @xmath1-algebra if and only if @xmath2 is weakly equivalent to a mapping space out of @xmath0 . in particular , spheres localized at a set of primes satisfy this condition .
in conclusion , we have found that the dynamical constraints brought by a strong interaction can trap the evolution of repulsive bosons hopping on a lattice into metastable states that lack translational symmetry , provided that the energy stored into the initial state is above a threshold . we pointed out that a self - induced effective attraction among doublons is one of the major processes that can effectively freeze the dynamics on long time scales . such a mechanism is recognized to play a role in the density relaxation processes of purely homogeneous systems through a dynamical arrest visible in time - dependent density correlations . the main features of this intriguing behavior , namely the slowing down of density excitations and the long - lived inhomogeneous pattern , resembles closely a kind of glass transition . moreover , we have shown that the time evolution of the many - body problem can be mapped onto that of a particle moving from the edge of a semi - infinite tight - binding chain with nearest - neighbor hopping , where each site represents a many - body wave function . this model looks like an anderson model , since both the on - site energy and the hopping vary from site to site , with a potential well at one edge due to the high - energy content of the initial state . interestingly , we find a delocalization - localization transition in this problem , with the particle being unable to diffuse on the whole chain above a certain value of the well depth . we consider this analogy quite suggestive and potentially constitute an even stronger indication of ergodicity breaking in the many - body space which is worth to be further investigated . we acknowledge discussions with g. biroli , j.f . carrasquilla , d. huse , a. parola , e. tosatti , f. zamponi , and c. kollath also for providing us with t - dmrg numerical data . computational support from caspur through the standard hpc grant 2011 is also acknowledged . all authors conceived and designed the research , equally contributing to the preparation of the manuscript . g.c . also designed and carried out the numerical calculations . the authors declare no competing financial interests . 10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , & _ _ ( , , ) . . _ _ * * , ( ) . , & . _ _ * * , ( ) . & . _ _ * * , ( ) . & . _ _ * * , ( ) . & . _ _ * * , ( ) . , & . _ _ * * , ( ) . , & . _ _ * * , ( ) . , , & . _ _ * * , ( ) . , & . _ _ * * , ( ) . . _ _ * * ( ) . , , , & . _ _ * * , ( ) . & . _ _ ( ) . _ et al . _ . _ _ ( ) . . _ _ * * , ( ) . & . _ _ * * , ( ) . . / downloads / software / tebd/. _ _ * * , ( ) . _ _ * * , ( ) . , , & . _ _ * * , ( ) . , , & & . _ _ * * , ( ) . . _ _ * * , ( ) . , , , & . _ _ * * , ( ) . & . _ _ * * , ( ) . & . _ _ * * , ( ) . , & . _ _ * * , ( ) . _ _ * * , ( ) . _ _ * * , ( ) . . _ _ * * , ( ) . , & . _ _ * * , ( ) . _ * * , ( ) . _ et al . _ . _ _ ( ) . , , & . _ _ * * , ( ) . _ _ et al . _ . _ the exact time evolution of an initial quantum state @xmath68 is given by : @xmath69 where @xmath70 are the eigenstates of @xmath45 , with corresponding eigenvalues @xmath71 , and @xmath72 . therefore , in order to follow the quantum dynamics to large times , the full knowledge of the spectrum is in principle needed . however , whenever the hilbert space is so large that it does not allow for a full diagonalization , an alternative approach is in order . at each time , we can make an evolution for a time step @xmath73 by considering a truncated taylor expansion of the unitary evolution : @xmath74 here , the series converges very quickly with @xmath75 and the cutoff @xmath76 must be chosen to obtain the desirable convergence for a given @xmath73 . therefore , the evolved wave function @xmath77 can be easily obtained by summing terms that can be in turn recovered by repeatedly applying the hamiltonian @xmath45 to @xmath78 . the full time evolution at long times can be achieved by subsequent small - time evolutions . moreover , time - dependent correlation functions of the form @xmath79 , where @xmath80 is an arbitrary operator , can be as well obtained by first applying @xmath81 to @xmath82 , then performing the time evolution @xmath83 , and finally computing @xmath84 , where @xmath85 . the numerical accuracy of the unitary evolution can be verified by checking the conserved quantities of the time - evolution ( for example the total energy ) which in our calculations remain to all purposes constant up to the longest considered evolution times . in this work we have used the lanczos method to construct a basis of @xmath86 orthonormal states in which the full hamiltonian takes a reduced tridiagonal form : @xmath87 the basis set is constructed in a recursive way starting from an initial vector @xmath88 upon repeatedly applying the hamiltonian @xmath89 with the coefficients of the tridiagonal matrix given by @xmath90 once the tridiagonal matrix is constructed we are in condition to consider a time - dependent problem of a particle initially localized on site @xmath88 whose dynamics is governed by @xmath91 . a final important remark concerns the practical numerical evaluation of the lanczos states and matrix elements on calculators with a finite - precision arithmetic . numerical truncation errors can indeed accumulate and lead the most recently generated lanczos vector not to be orthogonal to the previously generated ones . to overcome this problem , we have carefully analyzed the dependence of the results on the machine precision , considering up to @xmath92 bits of floating point arithmetic , for which no significant truncation error manifests up to the largest value of @xmath86 considered in the paper . the general idea of the time - evolving block decimation algorithm @xcite is to consider a particular representation of a generic many - body state defined in a hilbert space of dimension @xmath93 , where @xmath94 is the dimension of a certain local - basis . such a generic state is fully described in terms of the local quantum numbers @xmath95 as @xmath96 are taken to be @xmath97}\gamma_{\alpha_{0}\alpha_{1}}^{[1]i_{1}}\lambda_{\alpha_{1}}^{[2]}\gamma_{\alpha_{1}\alpha_{2}}^{[2]i_{2}}\lambda_{\alpha_{2}}^{[3]}\gamma_{\alpha_{2}\alpha_{3}}^{[3]i_{3}}\lambda_{\alpha_{3}}^{[4]}\cdot .. \cdot\lambda_{\alpha_{n-1}}^{[n]}\gamma_{\alpha_{n-1}\alpha_{n}}^{[n]i_{n}}\lambda_{\alpha_{n}}^{[n+1]},\label{eq : cvidal}\end{aligned}\ ] ] where the local tensors @xmath98 and the _ bond _ matrices @xmath99 constitute a set of @xmath100 parameters that specify the state in this representation . the time evolution of a given initial state is obtained upon considering a small time - step @xmath73 , and a repeated application of the real - time propagator @xmath101 within a suzuki - trotter approximation scheme , inducing systematic errors of magnitude @xmath102 . when the infinitesimal unitary evolution is applied the decomposition is consistently updated . the other major source of systematic errors is however intrinsically due to the finite amount of entanglement retained by vidal s representation . we have checked the convergence of the studied properties with respect to both the time step and the entanglement cutoff parameter , considering a maximum of @xmath103 , for which the density is converged up to the plotted times of fig . 2 ( right panel ) in the main paper . in this part of the supplementary materials , we give the details of the real - time variational monte carlo approach . first of all , we would like to stress that this method does not suffer from intrinsic instabilities arising from the sign ( or phase ) problem . indeed , our method is essentially a straightforward generalization of the variational technique , where the probability density is strictly non - negative . although approximated , this approach has firm physical basis and can accurately reproduce numerically exact results in cases where t - dmrg method can be applied ( e.g. , mostly in one - dimensional lattices ) . nonetheless , variational monte carlo can be generalized and applied in larger dimensions without any further numerical instability . in general , it is reasonable to expect that the relevant physical properties driving the dynamics can be reduced to a number of parameters significantly smaller than the size of the full hilbert space of the problem , as much as it is done in renormalization - based numerical and analytical tools . let us introduce a set of `` excitation operators '' @xmath104 , which are expected to describe the relevant excitations over the ground state that are involved in the dynamics of the system . the actual set of these operators has to be carefully chosen , depending upon the properties of the quantum phase . in the case of one - dimensional interacting bosons , @xmath104 can be selected to be density - density correlations , e.g. , @xmath105 , or doublon - doublon , holon - holon , and doublon - holon correlations . as discussed in the text , the presence of the former ones is sufficient to obtain a very accurate time dynamics for small quenches , whereas the inclusion of the latter ones is necessary for large quenches . in fig . [ fig : comparison ] , we report the comparison of the t - dmrg dynamics with our variational monte carlo approach . time - dependent expectation values of the on - site potential energy @xmath55 . the initial state is the ground state of the bose - hubbard hamiltonian with @xmath56 . real - time variational monte carlo results are obtained for an @xmath106 chain with periodic boundaries . corresponding data by t - dmrg @xcite are the potential energy at the center of an @xmath107 chain with open boundaries . ] let us discuss the details of our method . first of all , we define a basis @xmath108 in which the occupation numbers of the bosons are defined in the lattice . since the operators @xmath104 are diagonal in this basis , we will define @xmath109 as the ( time - independent ) value that the operator takes over the configuration @xmath108 . the variational wave function is given by : @xmath110\xi(\mathbf{x}),\label{eq : psivar}\ ] ] where @xmath111 is a time - independent uncorrelated state , such as the bosonic condensate , whereas @xmath112 and @xmath113 are time - dependent complex parameters . the variational form introduced here allows us to map the quantum dynamics associated to a given hamiltonian @xmath45 into the dynamics of the variational parameters @xmath112 and @xmath113 . the operators @xmath104 can be thought as collective variables for the quantum dynamics ; on physical grounds , we expect that the quantum dynamics will be characterized by a set of collective variables whose number @xmath114 is much smaller than the original size of the hilbert space . this reduction assumption is the key requirement for having an efficient time - dependent variational approach . the equations of motion for the variational parameters can be found by minimizing the euclidean distance in the hilbert space between the _ exact _ and the variational time dynamics . given at a certain time the quantum state @xmath115 , its exact time evolution leads to @xmath116 $ ] , where the complex - valued `` local - energy '' is @xmath117 . on the other hand , an infinitesimal change in time in the variational parameters leads to @xmath118 $ ] . the minimization of their mutual distance @xmath119 gives a set of coupled differential equations for @xmath120 and @xmath121 : @xmath122 where @xmath123 and @xmath124 denote expectation values over the square modulus of the wave function at time @xmath35 , namely @xmath125 . the coupled equations of motion determine a consistent hamiltonian dynamics in which at all times both the norm @xmath126 and the expectation value of the energy @xmath127 are strictly conserved . indeed , one can show that @xmath128 $ ] and @xmath129 , which are both vanishing for the optimal solutions for which @xmath130 is at a minimum . not surprisingly , the same equation of motions that we have derived by means of an optimal distance principle can be also derived from the variational principle , namely it can been shown that the action @xmath131 is indeed stationary for the time - trajectories of the variational parameters induced by eqs . and . the time - dependent expectation values over the square modulus of the many - body wave function can not be calculated analytically for generic operators @xmath104 , and a variational quantum monte carlo is needed to evaluate them . in particular , @xmath132 can be straightforwardly interpreted as a probability distribution over the hilbert space spanned by the configurations @xmath133 and a markov process can be devised , such that the stationary equilibrium distribution coincides with the desired probability measure . in the main paper , we have concentrated our attention on the unitary - filling density , where at equilibrium a mott transition takes place upon increasing the interaction strength . to better realize the importance of the density on our considerations on the large-@xmath2 relaxation times , we now show the exact time evolution of inhomogeneous states at @xmath37 . for example , we consider @xmath38 and an initial density profile @xmath39 . as shown in fig . [ fig : neq1 ] , we find quite a different behavior for the density relaxation times with respect to the unitary - filling case , with a much smoother crossover from small to large values of @xmath48 . moreover , there is no evidence of any increase in the relaxation times with the system size . this fact suggests that the dynamical constraints brought by the effective interaction among doblons are much stronger at @xmath14 , where a sharp crossover in the relaxation times is observed . we finally remark that , even at non unitary fillings , we expect larger clusters of doblons to have larger relaxation times than the simple initial state we have considered here . the possibility of a non - thermal behavior at non - unitary filling can not be excluded , even though the strongest manifestation of the dynamical arrest is expected to happen at @xmath14 . exact time evolution of the densities on initially doubly occupied sites at filling @xmath38 for an @xmath135 chain . different curves from left to right correspond to increasing values of the final interaction at @xmath136 . inset : corresponding inverse relaxation times @xmath137 for increasing lattice size @xmath138 , @xmath139 , and @xmath140 . ] to better elucidate the many - body localization phenomenon suggested by the lanczos - basis analysis of the bose - hubbard model , we hereby show that localization is _ not _ strictly implied by integrability . to this purpose , we consider a model of one - dimensional hard - core bosons prepared in an initial inhomogeneous state with an alternating density profile @xmath10 and let it evolve in the lanczos basis with their non - interacting ( integrable ) kinetic hamiltonian . in fig . [ fig : v0mod ] , we show both the lanczos hopping elements and the expectation value of the lanczos particle traveling in the many - body space . the effect of integrability is to reduce the number of allowed lanczos states with respect to the total hilbert space . indeed , we find that the hopping is a strongly decreasing function of the iterations , approaching zero after a certain number of states @xmath141 that is a finite fraction of the full hilbert space . noticeably , the average position traveled by the particle increases accordingly , leading , therefore , to a full delocalization in the thermodynamic limit . this analysis shows that a localization in the lanczos basis is not necessarily due to integrability whose effect only amounts to reduce the number of active many - body states whereas it is due to the effective dynamical constraints brought by the interaction .	when classical systems fail to explore their entire configurational space , intriguing macroscopic phenomena like aging and glass formation may emerge . also closed quanto - mechanical systems may stop wandering freely around the whole hilbert space , even if they are initially prepared into a macroscopically large combination of eigenstates . here , we report numerical evidences that the dynamics of strongly interacting lattice bosons driven sufficiently far from equilibrium can be trapped into extremely long - lived inhomogeneous metastable states . the slowing down of incoherent density excitations above a threshold energy , much reminiscent of a dynamical arrest on the verge of a glass transition , is identified as the key feature of this phenomenon . we argue that the resulting long - lived inhomogeneities are responsible for the lack of thermalization observed in large systems . such a rich phenomenology could be experimentally uncovered upon probing the out - of - equilibrium dynamics of conveniently prepared quantum states of trapped cold atoms which we hereby suggest . the ergodicity axiom in classical statistical mechanics states that , during its time evolution , a closed macroscopic system uniformly explores the entire phase space compatible with conservation laws , so that the time average of any observable comes to coincide with the micro - canonical ensemble average and , when the observable is local , also with the canonical gibbs ensemble average . nonetheless , ergodicity can be violated in classical systems , a noticeable example being glasses . @xcite quantum effects might also spoil ergodicity by preventing the wave function from diffusing within all available configurations . this phenomenon is actually known to occur in the presence of disorder and manifests itself either by single - particle @xcite or many - particle @xcite wave function localization . however , alike classical models for glassy behavior , ergodicity breakdown in the quantum dynamics may not necessarily require disorder and it could instead be entirely due to frustrating dynamical constraints . @xcite this issue is currently attracting great interest , @xcite since well controlled realizations of closed quantum systems have become feasible upon trapping cold atomic species . @xcite indeed , similarly to what can be done in numerical simulations , one can prepare atoms in a given initial state and probe their time evolution under a hamiltonian whose parameters are fully under control , thus offering the unique opportunity to monitor the ergodicity principle at work in the quantum realm . @xcite in this work , we report numerical evidences that an isolated system of strongly interacting bosons , modeling atoms in optical lattices , can be trapped during its evolution into long - lived inhomogeneous metastable states , provided that its internal energy exceeds a certain threshold . we argue that the slowing down of high - energy incoherent excitations in the strongly correlated system is the key feature responsible for this dynamical arrest , much resembling a kind of glass transition . by formulating the problem in a different language , we explicitly show that a system initially prepared in a inhomogeneous state is unable to diffuse within the entire configurational space ; such a dynamical localization in the many - body hilbert space looks intriguing and may represent a kind of many - body anderson localization @xcite that occurs without disorder . the above phenomenon is put in further relation with and deemed responsible for the lack of ergodicity observed in large finite size systems . this belief is confirmed by means of a novel time - dependent variational monte carlo method that we introduce hereby and an experimental set up to uncover this very rich phenomenology is also suggested . one of the simplest models that can be realized in experiments is the bose - hubbard hamiltonian : @xcite @xmath0 characterized by the amplitude @xmath1 for a bosonic species of an atom to hop between nearest - neighboring wells of an optical lattice and by a _ local _ repulsion @xmath2 among atoms localized in the same potential well . the operators @xmath3 and @xmath4 create and destroy , respectively , a boson on site @xmath5 , and @xmath6 is the density operator . experiments are often performed with anisotropic lattices that realize a collection of almost uncoupled chains , a fortunate case for numerical simulations that we shall mainly consider hereafter , apart from a brief excursion in two dimensions towards the end of the paper . in one dimension , obstacles to ergodicity can arise in integrable models.@xcite however , the hamiltonian is not integrable and , indeed , there are experimental evidences that its dynamical evolution may succeed in fast relaxing to a _ thermal _ state . specifically , in a recent experiment a system of @xmath7rb atoms , well described by the hamiltonian , has been prepared in a state in which the sites of the optical lattice were alternatively empty and singly occupied . @xcite this state was let evolve for different experimental conditions corresponding to different ratios @xmath8 . even at the largest value @xmath9 , the initial density profile @xmath10 was found to rapidly relax to the homogeneous _ thermal _ one @xmath11 , with half a boson per site , much faster than the integrable counterparts of non - interacting or infinitely - interacting ( i.e. , hard - core ) bosons and consistently with the increased number of relaxation channels that opens once integrability is lost . numerical simulations of the above experiment successfully reproduce the observed _ thermal _ behavior . @xcite however , there are also numerical evidences pointing out a breakdown of ergodicity in the same model but within a different region of the parameter space , specifically when the number of bosons is one per site . this case , as well as any other one with integer density , is special because , at equilibrium and at zero temperature , the model undergoes a quantum phase transition into a mott insulator above a critical @xmath8 . even though the phase transition is washed out by thermal fluctuations , nonetheless its influence on the spectrum seems to prevent thermalization above a certain @xmath8 in numerical simulations of finite size systems@xcite ; a result that may @xcite or may not @xcite prelude to a true breakdown of ergodicity in the actual thermodynamic limit . here , we propose an experiment a lot alike the one previously described , @xcite which may distinguish very sharply a change from chaotic to non - chaotic dynamics in the bose - hubbard model . our proposal starts by observing that , in the gapless phase next to the mott insulator at one boson per site , low - energy itinerant bogoliubov quasiparticles should coexist with high - energy incoherent excitations , which , for @xmath12 , can be identified as sites occupied by more than a single boson . it is well possible that the overall relaxation depends critically upon the population of those high - energy excitations , @xcite which can be assessed by tailoring initial states with lots of doubly occupied sites rather than none as in the experiment of ref . @xcite . this is indeed what we propose and simulate by means of different and complementary numerical tools , such as exact diagonalization , time - evolving block decimation @xcite and a novel time - dependent variational monte carlo algorithm , as discussed in more detail in the supplementary material . inhomogeneous initial states constituted by clusters of doubly occupied sites ( large blue circles ) and empty sites ( small dots ) . the size of each cluster is denoted by @xmath13 . ] we start from analyzing the model at density @xmath14 , when at equilibrium a mott transition occurs at a critical @xmath15 . @xcite we imagine to prepare an initial state where all the sites are either empty or doubly occupied . in particular we consider the states depicted in fig . [ fig : deftype ] , namely with clusters of doubly occupied sites of variable size @xmath13 . these states are let evolve with the spatially homogeneous hamiltonian for different @xmath8 , below and above the critical value . while for small @xmath8 the density profile rapidly reaches the equilibrium configuration @xmath16 , for large @xmath8 , it stays close to its initial value for a remarkably long time . eventually , since the system is finite , the density profile approaches the homogeneous plateau , with small residual oscillations that get damped as the system size increases . we can define a relaxation time @xmath17 ( whose inverse is shown in fig . [ fig : decdef ] , for @xmath18 ) , as the first time for which the local density approaches its homogeneous value . we highlight that @xmath17 , at a specific @xmath19 , has a sudden step up , which becomes sharper and sharper as the system size increases . the above results show that above @xmath19 the system has the tendency to stay dynamically trapped into long - lived inhomogeneous configurations . we emphasize that such a behavior is all the more remarkable not because doubly occupied sites seem unable to decay , which is known to require for @xmath20 very long times and large system sizes , @xcite but rather because they are not capable to move , hence restore translational symmetry . we can better understand this surprising result by an effective hamiltonian that can be derived following the same reasoning of refs . @xcite . for a sufficiently large interaction , it is justified to project the evolution onto states with the same potential energy per site @xmath21 , at least for time scales shorter than @xmath22 one realizes that states with the same potential energy but with triply and singly occupied sites start to contribute only at order @xmath23 , so that , with accuracy @xmath24 , the number of doubly occupied sites is conserved . if we associate a fictitious spin up or down to a doublon ( doubly occupied site ) or a holon ( empty site ) , respectively , we find that the effective hamiltonian that controls the evolution reads : @xmath25,\label{eq : heff}\ ] ] which describes a hard - axis ferromagnetic heisenberg model . @xcite we note that the evolution of an xxz spin chain starting from an antiferromagnetic ordered initial state has been studied numerically in ref . @xcite with dmrg . here we consider the effective dynamics ( [ eq : heff ] ) in a regime when the anisotropy would favor a ferromagnetic ground state and when the initial state contains clusters of up and down spins of increasing lenght . the smaller cluster size of @xmath26 corresponds to a neel initial state considered in ref @xcite . in the right panel of fig . [ fig : decdef ] we show the time - dependence of the clusters density in the large @xmath2 regime , evolved according to the effective hamiltonian . remarkably , we see that even for small clusters the system fails to restore the spatial homogeneity up to very long time scales , which turn to be far beyond those currently accessible in typical experimental setups . the slowing down of the dynamics can be traced back to the effective attraction among doublons , @xcite which makes their aggregates hard to break up . in other words , what seems to matter more is the dissociation of clusters of doublons , rather than the decay of a single one . indeed we have checked ( not shown ) that in the large @xmath2 regime the system has the tendency to get dynamically stuck into clusters of doublons of finite size , whereas a much faster annihilation and recombination rate is observed in the small @xmath2 limit . to get further insights into the dynamical behavior of the system , we recast the problem in a different language . starting from the initial state , denoted as @xmath27 , we can generate an orthogonal basis set @xmath28 , @xmath29 , by repeatedly applying the hamiltonian ( see supplementary material ) . in this lanczos basis of many - body wave functions , the hamiltonian has the form of a tight - binding model on a semi - infinite chain . each site @xmath29 corresponds to a many - body state , it has an on - site energy @xmath30 and is coupled only to its nearest neighbors by hopping elements @xmath31 and @xmath32 . @xcite it is easy to realize that the unitary evolution of the original many body problem is thus fully equivalent to the dynamics of a single particle , initially sitting at site @xmath33 , that is then let propagate along such a tight - binding chain of many - body states . we note that , both @xmath34 and @xmath31 largely fluctuate from site to site , therefore resembling an effective anderson model , even though those parameters are in reality deterministic , see fig . [ fig:1d ] . in the same figure , we also show the mean distance traveled by the particle after time @xmath35 starting from the first site of the chain , which corresponds to an initial state @xmath36 , and for different @xmath8 . we observe that , for small values of @xmath8 , the particle diffuses and its wave - packet finally spreads over the whole chain , in a rather uniform way . on the contrary , above a certain critical value of the interaction @xmath19 , the particle stays localized near the origin for arbitrarily long times . a closer look to the structure of the on site energies reveal the existence of a potential well at the edge of the chain . this is crucial in order to understand the observed localization transition . since the potential well can not induce a true bound state below the bottom of the spectrum , at most a resonance will form in the spectrum , from which the particle could in principle escape in a finite time . this indeed happens at small @xmath2 but apparently not at large @xmath2 , where the increased depth of the well and the effective randomness of the on - site energies conspire together to keep the particle localized close to the edge , preventing the the states in the well to hybridize with other states along the chain . we now see how this result connects with the previous analysis on the density relaxation times . in the small @xmath8 regime the particle is able to escape from the well and to explore larger portions of the chain , thus resulting into a fast density relaxation . as opposite , for large @xmath8 , the particle bounces back and forth inside the well , finding hard time to escape from it . this lack of diffusion results into a very long - time scale for the density to relax to its homogeneous value . the above results show explicitly that some kind of localization in the many - body configurational space does occur , at least in the finite system . @xcite while such an intriguing behavior might well be a subtle effect due to the finite size spectrum , it could also signal the onset of a genuine localization that survives in the thermodynamic limit . we conclude the discussion on the inhomogeneous initial states by studying the case at density @xmath37 , when at equilibrium there is no longer a mott transition . for example , we consider @xmath38 and an initial density profile @xmath39 . interestingly , we find quite a different behavior for the density relaxation times @xmath17 ( see supplementary material ) , with a much smoother crossover from small to large values of @xmath8 and no evidence of any increase in the relaxation times with the system size . this fact suggests a deeper connection between the observed dynamical behavior and the zero temperature mott transition that occurs at equilibrium and at integer filling , as suggested by calculations with infinite - coordination lattices . @xcite inverse relaxation times of density excitations in the homogeneous system , see eq . . from left to right , different curves correspond to different initial states at @xmath40 , @xmath41 , and @xmath42 . insets : real part of the density correlations @xmath43 in the ergodic and in the non - ergodic region . data are obtained with exact - diagonalization on a lattice with @xmath44 . ] in light of the previous results , one may question that by choosing an inhomogeneous configuration of doublons we pick up a rather specific initial state in the hilbert space . we are now going to show that the above findings strongly affect the dynamics starting from a perfectly homogeneous state . in this respect , a particularly interesting class of initial states are the ground states of @xmath45 for given values of the interaction @xmath46 , which are let evolve under the hamiltonian dynamics after a sudden increase of the interaction to a final value @xmath47 , the so - called quantum quench . kollath and coworkers @xcite reported evidence for the existence of two separated regimes in which either thermal or non - thermal behavior is observed for local observables . the origin of the non - thermal behavior in the large @xmath48 region and the possibility of an ergodicity breaking in the thermodynamic limit is still highly debated @xcite . in the following , we focus on an average density @xmath14 and we show that signatures of long lived metastable states of doblons can be identified in the dynamics after a quantum quench . at variance with the previous numerical calculations , now both the initial state and the quantum hamiltonian do preserve the spatial homogeneity and , therefore , the quest for possible signatures of ergodicity breaking requires a different approach . since we have identified density relaxation as the slowest process in the problem , we monitor the dynamics of the system by measuring the auto - correlation of the density averaged over all sites , namely through @xmath49 for any finite size system , @xmath50 since the densities decorrelate at very long times . indeed , as shown in fig . [ fig : relax ] , for small @xmath51 this quantity has a very fast transient to zero . on the contrary , for @xmath52 the density auto - correlation @xmath43 gets stuck into a long - lived finite value plateau @xmath53 , before approaching zero only on a much longer time scale . if we extract a relaxation time from @xmath43 , we find a similar behavior as in fig . [ fig : decdef ] , i.e. , a dramatic increase of the relaxation times above a threshold value of the final interaction strength . in agreement with the previous analysis , the appearance of such a long - lived metastable state characterized by a finite plateau @xmath54 of the density auto - correlation function might indicate an excess of double occupancies that have no channel to relax . in other words , the dynamical constraints brought by the interaction severely slow down density excitations , whose characteristic time scales increase abruptly after a critical threshold . the main phenomenological traits of this _ dynamical arrest _ characterized by long - lived inhomogeneous states closely remind the physics of glassy materials . the previous discussion revealed the existence of a threshold energy above which a steep increase of the relaxation time of density fluctuations takes place . in order to assess the relevance of this phenomenon for the dynamics of larger systems or even for higher spatial dimensions , it is desirable to devise a comprehensive alternative framework able to catch its very characteristics . here , we introduce an approach based on real - time variational monte carlo ( see supplementary material for details ) , which has two important advantages : it allows us to follow the evolution for times comparable to those accessible experimentally , which are much longer than t - dmrg ; @xcite it can be easily extended to higher dimensions . we mention that mean - field - like variational approaches to the real - time dynamics of correlated systems have been developed in recent years . @xcite one - dimensional results for the time - dependent expectation values of the on - site potential energy @xmath55 in the ergodic ( _ upper panel _ ) and in the non - ergodic regions ( _ lower panel _ ) . the initial state is the ground state of the bose - hubbard hamiltonian with @xmath56 and the considered system size is @xmath57 . grand - canonical thermal averages are shown for comparison as dashed horizontal lines . ] however , although they seem to capture well the main features of the dynamical evolution , these methods are unable to describe important aspects such as damping and relaxation . on the contrary , our approach is sufficiently rich to account for damping and relaxation of local observables . it is based on a very simple and transparent out - of - equilibrium extension of the jastrow - like variational wave function that was shown to describe quite accurately the equilibrium phase diagram of the bose - hubbard model : @xcite @xmath58 where @xmath59 is the initial state and @xmath60 is a jastrow factor that depends on the occupancies @xmath61 and @xmath62 of two sites @xmath5 and @xmath63 and varies with time so to maintain the time evolution as close as possible to the true evolution via the schroedinger equation ( see supplementary material ) . the comparison between our approach and the t - dmrg @xcite is reported in the supplementary materials , demonstrating the high accuracy of the time - dependent variational monte carlo . results for the time evolution of a local observable , such as the potential energy , after a sudden quench from @xmath56 to a final @xmath48 are shown in fig . [ fig : eq - values ] . the values of the thermal averages have been computed in the grand - canonical ensemble by means of finite - temperature quantum monte carlo calculations , @xcite with the effective temperature fixed by the average energy of the initial state , which we take as the best variational approximation for the ground state at @xmath56 . as shown in fig . [ fig : eq - values ] , in the region of small @xmath48 we observe a damping of the average potential - energy , which approaches a quasi - steady stationary value in contrast to the simple gutzwiller wave function , @xcite . in this regime , damping is mainly due to a density - density jastrow factor of the form @xmath64 , which already at equilibrium was shown to provide a satisfactory description of the physical behavior . @xcite this fact enlightens the relevance of the bogoliubov modes whose dephasing during the time evolution allows to approach the stationary state . remarkably , the steady state averages coincide with the thermal ones ; a signal that the dynamics is ergodic . in the region of large interactions @xmath48 , a simple density - density jastrow factor does not account for all relaxation pathways , which will now mainly result from specific correlations among doublons , holons and between holons and doublons . the effective hamiltonian indeed explicitly shows that doublons attract each other as well as holons do , while doublons repel holons . these correlations , as well as other among higher on - site densities , can be easily implemented via the the jastrow factor in eq . and indeed substantially improve the dynamics . interestingly , the effective interaction between doblons that results from the dynamical variational calculation turns to be _ attractive _ , therefore leading to a consistent determination of the anticipated dynamical effects that drive the dynamics in this regime . as we see from fig . [ fig : eq - values ] , in the region of very large @xmath48 the potential - energy expectation values do show a damping to a non - thermal quasi - steady state on a time of the order of @xmath65 . this fast time scale must be put in comparison with the much longer one , @xmath17 of fig . [ fig : relax ] . it is natural to identify @xmath17 with the time scale that controls the eventual escape from the quasi - steady state , hence the approach to thermal equilibrium . whether this time scale does truly diverge in the thermodynamic limit , or rather saturates to a very large value which is still out of reach for state of the art numerics , it is certainly an important issue that can not be definitively solved . however , we can safely state that a large finite system of actual experimental relevance will get stuck for a quite long time into highly inhomogeneous metastable states , which we revealed to be on the verge of a spatial symmetry breaking . the above results have been obtained for out - of - equilibrium one dimensional systems and therefore leave the questions concerning the dependence on the dimensionality of the problem still open . however , the anomalously long - time relaxation of the density auto - correlation points towards a kind of glassy behavior that should be observable even in higher dimensions , provided that the interaction induces sufficiently strong dynamical constraints . in this regard , we have studied the two - dimensional case by means of our time - dependent variational scheme , verifying that a similar behavior occurs even in two dimensions . in fig . [ fig : eq - values-2d ] , we show the results of the potential - energy expectation values as a function of the time . as before , if the repulsion is weak , the time average coincides with the thermal ones , while for strong repulsion there is a clear difference between thermal and time averages , giving rise to a nonergodic dynamics . we therefore argue that the phenomenology we have hereby identified is almost independent on the dimensionality and it is rather due to the existence in strongly correlated systems of high - energy incoherent excitations that do not have channels to relax efficiently . this scenario is also consistent with the observed strong doping dependence of the relaxation time ( see supplementary material ) . two dimensional results for the time - dependent expectation values of the on - site potential energy @xmath55 in the ergodic ( _ upper panel _ ) and in the non - ergodic regions ( _ lower panel _ ) . the initial state is the ground state of the bose - hubbard hamiltonian with @xmath66 and the considered system size is @xmath67 . grand - canonical thermal averages are shown for comparison as dashed horizontal lines . ]
in the @xmath1cdm cosmological scenario about 80@xmath2 of the matter of the universe is believed to be composed of non - baryonic matter , called dark matter ( dm ) . the most popular dm candidates are the wimps ( weakly interacting massive particles ) supposed to be cold , electrically neutral , stable , and massive @xcite . among the huge plethora of wimp candidates , the best motivated ones are related to the supersymmetrical ( susy ) and extra dimensional extensions of the standard model of particle physics @xcite . + in the minimal supersymmetric extension of the standard model ( mssm ) , the neutralino @xmath3 represents an excellent cold dm candidate with a relic density compatible with the wmap bounds . since the neutralino is a majorana particle , pairs of @xmath3 can annihilate into standard model particles , e.g. , quarks , leptons , and w bosons . the subsequent hadronization of those particles results in a continuum emission of gamma - rays characterized by a cut - off at the neutralino mass and by possible spectral features like bumps or a hardening of the spectral slope . + the expected gamma - ray flux from dm - annihilating astrophysical objects , as function of the energy threshold @xmath4 and the integration region @xmath5 , within which the signal is integrated , can be factorized in two terms : @xmath6 the so - called particle physics factor @xmath7 depends on the features of the dm particle , and can be written as : @xmath8 where @xmath9 is the velocity averaged annihilation cross - section , and @xmath10 is the particular branching ratio for the @xmath11-th annihilation channel . + the term j@xmath12 ( the so - called astrophysical factor ) is given by the line of sight integral over the dm density squared within a solid angle @xmath5 , and depends on the density profile of the dm halo of the source : @xmath13 + since the gamma - ray flux of dm annihilation is proportional to the square of the dm density , only sources with high expected dm densities are good targets for dm indirect searches . among these , the dwarf spheroidal satellite galaxies ( dsphs ) of the milky way ( mw ) are interesting objects thanks to their relative proximity to the earth , to their high mass to light ratio ( with values within tens and thousands of m@xmath14/l@xmath14 ) and to the expected absence of conventional gamma - ray sources within the system @xcite . so far , around two dozen dsphs have been identified . segue 1 , discovered in 2006 by the sdss @xcite , is located at 28 kpc from the galactic center , at ( ra , dec)=(10.12@xmath15 , 16.08@xmath16 ) . kinematics studies applied to 66 member stars allowed to estimate its mass to light ratio to be in the range 1320 - 3400 m@xmath14/l@xmath14 @xcite , highlighting segue 1 as the most dm dominated dsph known so far . + the magic - i telescope is a 17 m dish imaging atmospheric cherenkov telescope ( iact ) , located at the roque de los muchachos observatory , in the canary island of la palma ( 2200 m a.s.l . ) . thanks to its low energy threshold ( @xmath1760 gev at zenith ) , high flux sensitivity ( @xmath18 of the crab nebula flux in 50 hour of observations above @xmath17250 gev ) , and good angular and energy resolution ( @xmath19 and @xmath20 respectively , at @xmath0 gev ) @xcite , magic - i is a suited instrument for the indirect search for dm candidates with energy of the neutralino mass or the kaluza - klein state . a search for a possible dm gamma - ray signal coming from segue 1 was performed by the magic - i telescope between november 2008 and march 2009 , for a total of 29.4 hours of observation time ( after data selection ) . the data analysis was performed using the standard magic - i analysis and reconstruction software @xcite . the number of gamma - ray candidates events from the direction of the source was estimated using the distribution of @xmath21 angles , which are related to the orientation of the showers . the overall analysis cuts were optimized and cross - checked for point - like sources with the aid of contemporaneous crab nebula data . in figure [ alphaplot ] , the @xmath21-plot above @xmath0 gev is shown . the number of excess events was computed in a fixed fiducial signal region with @xmath22 and resulted to be @xmath23 , corresponding to a significance of @xmath24 , computed using eq.(17 ) of li@xmath25ma @xcite . -plot from 29.4 hours of segue 1 observation above @xmath0 gev . red points represent the signal ( on distribution ) , black points the background ( off distribution ) , and green points their difference . the vertical dashed line at @xmath26 is the fiducial region below which the excess event number is estimated . image taken from @xcite.,width=264 ] since results were consistent with no signal over the background , we derived upper limits ( uls ) on the flux , calculated using the rolke method @xcite at @xmath27 confidence level , and assuming a @xmath20 systematic uncertainty . figure [ intul ] shows the integral uls achieved by the magic - i observation of segue 1 considering different energy thresholds @xmath28 and different power - law spectra with spectral index @xmath29 . is assumed . image taken from @xcite.,width=264 ] it is worth noting that , using the rolke method , the uls on the number of the excess events , and consequently the integral flux uls , are affected by statistical fluctuations quantified by the significance of the observation @xmath30 . this is an intrinsic feature of the statistical method exploited in the analysis and it should be taken into account when comparing uls from different analyses . to show this effect , in figure [ intul ] we plot also the uls ( dashed lines ) computed assuming a value for @xmath30 equal to zero ( with number of on events equal to the number of off events in the signal region of @xmath21-plot ) for different values of spectral index and energy threshold . assuming a particular form for segue 1 dm halo , and a given particle model for the dm candidates , we can translate the integral uls derived from the segue 1 observation into constraints on the dm annihilation rate . + motivated by results from cosmological simulation , the dm halo around segue 1 was modeled by using the einasto radial profile @xcite with @xmath311.1@xmath3210@xmath33 m@xmath14 kpc@xmath34 , r@xmath35=0.15 kpc , and n=3.3 . with those parameters , the total astrophysical factor of segue 1 results to be @xmath3610@xmath37 gev@xmath38 @xmath39 sr . since the analysis was performed assuming point - like source cuts ( corresponding to an angular integration of 0.14@xmath40 above @xmath0 gev ) , we estimated the effective astrophysical factor within the analysis cuts to be used in the following analysis , being its value @xmath4110@xmath37 gev@xmath38 @xmath39 sr ( corresponding to the 64@xmath2 of the total astrophysical factor ) . + concerning the particle physics , we restricted ourselves to the case of a susy model in which the presence of a discrete symmetry ( r parity ) guarantees that the lightest supersymmetric particle ( lsp ) is stable over cosmological timescales and , therefore , a good dm candidate . we considered a 5-dimensional subspace of the mssm called msugra @xcite , for which the basic parameters are the universal masses of the gauginos ( @xmath42 ) and scalars ( @xmath43 ) , the trilinear coupling ( @xmath44 ) , the ratio of the vacuum expectation values of the two higgs fields ( tan@xmath45 ) and the sign of the higgsino mass term ( sign@xmath46 ) . in order to study the phenomenology of msugra we performed a grid scan over the parameter space , for a total of 5@xmath3210@xmath47 points ( for the details see @xcite ) . + the full circles of figure [ dm_uls ] represent all the models of the scan _ i ) _ where the lightest susy particle is a neutralino , _ ii ) _ that survive the standard model constraints and _ iii ) _ with a relic density compatible with the value derived by wmap data within three times its experimental error @xmath48 @xcite . for each dm model of the scan , we computed the integral flux ul ( above an energy threshold e@xmath49 ) , using the segue 1 data and the specific gamma - ray spectrum derived from the individual dm model . since the spectra for each dm model have different shapes and cut - offs , the value of the optimal energy threshold e@xmath49 was computed individually for each dm mass . we then converted the flux uls into uls on the velocity averaged cross - section to have a direct comparison of experimental data with the theoretical predictions . the results are plotted in figure [ dm_uls ] as function of the neutralino mass : each dm model of the scan ( full circle ) is compared to its own ul ( square ) . for each point we defined an enhancement factor ( enf ) as the ratio between the ul on the velocity averaged cross - section and the value predicted by msugra . this quantity quantifies how far away we are from excluding some portions of the msugra parameters space . from figure [ dm_uls ] it can be seen that enfs for model compatible with the wmap bounds are typically above 10@xmath50 , while typical values are of the order of @xmath51 . from wmap bounds . for each of these full circles the ul on the cross - section can be computed from the segue 1 data ( after energy threshold optimization ) and it is indicated here by a square . circles and squared are colored in term of the enhancement factor . image taken from @xcite.,width=264 ] we tested our uls on some of the models proposed in the literature that can explain the pamela data @xcite for the energy spectrum of the positron fraction @xmath52 as due to dm annihilation into leptons . the regions in the @xmath53 plane that provide a good fit to the pamela measurements @xcite for three different channels of dm annihilation are shown in figure [ pamela_uls ] . the annihilation channels @xmath54 , @xmath55 have been taken from supersymmetry , while for @xmath56 the existence of a new dark force , mediated by the carrier @xmath57 that decays into leptons @xcite , has been assumed . in figure [ pamela_uls ] we plot the uls obtained from the segue 1 data , using again the specific dm annihilation spectra . we can see that , in this case , the enfs needed to meet the pamela - favoured region are much smaller than those found for msugra , and in the case of annihilation into @xmath58 our uls are probing the relevant regions . however , it is worth mentioning that , since the uncertainty in the segue 1 astrophysical factor is quite large @xcite , an improvement in accuracy of astrophysical factor value could be able to put more stringent constrains and to confirm the exclusion of the pamela region for dm particle annihilating in @xmath58 . ( green lines ) or @xmath58 ( blue line ) , and for a dm candidate interacting with a light intermediate state @xmath57 decaying into a pair of electrons ( pink line ) . the same annihilation channels ( with the same color coding ) are considered to draw the regions in the plane that provide a good fit to the pamela measurement of the energy spectrum of the positron fraction . image taken from @xcite.,width=264 ] a search for a possible dm gamma - ray signal coming from segue 1 was performed by the magic - i telescope . no hints of signal were found above the background for energies larger than @xmath0 gev . integral uls on the gamma - ray emission were computed assuming different power - law energy spectra . within the msugra scenario , a large scan of neutralino models was performed over the parameter space . subsequently for each simulated dm model , the uls on the velocity averaged annihilation cross - section @xmath59 were derived separately for each point in the scan in order to account for the dependence on the specific spectra . results indicate that a general exclusion plot can not be drawn to constrain the parameter space , so we provide the results in terms of enhancement factors . a minimum boost on the flux is found of the order of 10@xmath60 ( for models compatible with wmap ) while the typical values are at 10@xmath61 . magic - i data of segue 1 can be useful to put constraints on those dm models that are provided in the literature to explain the pamela data . our uls are probing the pamela region for the dm models annihilating into @xmath58 but the robustness of this result could be improved , decreasing the uncertainty in the astrophysical factor . + although the magic - i observation did not result in a detection , and the uls require still high flux enhancement factors to actually match the experiment sensitivity , an analysis like the one presented here is able to point out details and features that can be important for future deep exposures of this or similar objects , with next - generation cherenkov experiments . 9 d. n. spergel _ et al . _ , apjs * 170 * ( 2007 ) 377 . g. bertone , d. hooper , j. silk , phys . rept . * 405 * ( 2005 ) 279 - 390 . m. sanchez - conde _ et al . _ , aip conf . proc.1166 ( 2009 ) 191 - 196 . g. gilmore _ _ , astrophys.j.663 ( 2007 ) 948 - 959 . v. belokurov _ et al . _ [ sdss colaboration ] , astrophys.j . * 654 * ( 2007 ) 897 - 906 [ astro - ph/0608448 ] . simon _ et al . _ , ( submitted ) ( 2010 ) astro-ph.ga/1007.4198 . j. albert _ et al . _ [ magic collaboration ] , astrophys.j . * 674 * ( 2008 ) 1037 - 1055 [ arxiv:0705:3244 ] . t. bretz _ et al . 29th icrc 2005 , tsukuba . li , y.q . ma , astrophys.j . * 272 * ( 1983 ) 317 - 324 . j. aleksi _ et al . _ [ magic collaboration ] , jcap * 06 * ( 2011 ) 035 . w. a. rolke , a. m. lopez , j. conrad , nucl.instrum.meth . * a551 * ( 2005 ) 493 - 503 [ physics/0403059 ] . r. essig , n. sehgal , l. e. stringari , m. geha , j.d . simon , phys.rev * d82 * ( 2010 ) 123503 [ astro - ph/1007.4199 ] . a. h. chamseddine , r. l. arnowitt , p. nath , phys.rev.lett * 49 * ( 1982 ) 970 . e. komatsu _ et al . _ [ wmap collaboration ] , astrophys.j.suppl . * 192 * ( 2011 ) 18 [ astro-ph.co:1001.4538 ] . o. adriani _ et al . _ [ pamela collaboration ] , nature * 458 * ( 2009 ) 607 - 609 [ astro - ph/0810 - 4995 ] . i. cholis _ et al . _ , jcap * 0912 * ( 2009 ) 007 [ astro - ph/0810.5344 ] . i. cholis _ et al . _ , phys.rev * d80 * ( 2009 ) 123518 [ astro - ph/0811.3641 ] . n. arkani - hamed _ et al . _ phys.rev * d79 * ( 2009 ) 015014 [ hep - ph/0810.0713 ] .	despite the interest in dark matter ( dm ) searches is currently more focused on underground experiments , a signature of dm annihilation / decay in gamma - rays from the space would constitute a smoking gun for its identification . here , we present the results of the survey of segue 1 by the magic - i telescope performed in 2008 and 2009 . this source is considered by many as the most dm dominated milky way satellite galaxy known so far . the nearly 43 hours of data taken constitute the deepest observation ever made on a single dwarf galaxy by cherenkov telescopes . no significant gamma - ray emission was found above an energy threshold of @xmath0 gev . integral upper limits on the gamma - ray flux were calculated assuming various power - law spectra for the possible emission spectrum and for different energy thresholds . we also discuss a novel analysis that fully takes into account the spectral features of the gamma - ray spectrum of specific dm models in a supersymmetric scenario .
minimizing hypercones have fascinating originations and active developments in geometric measure theory , for example , the pioneering works fleming @xcite , de giorgi @xcite , almgren @xcite , simons @xcite , bombieri - de giorgi - giusti @xcite for solving the celebrated bernstein problem . many types of minimizing hypercones beyond simons cones were discovered by lawson in @xcite . later @xcite added that @xmath0 over minimal clifford torus @xmath1 is minimizing ( also see @xcite ) whereas @xmath2 is only stably minimal . although lots of examples had been found , a complete list of homogeneous minimizing hypercones remained unknown until the important work of lawlor @xcite . rather than in the whole euclidean space , lawlor searched for a similar bundle structure to the characterization foliation of a minimizing hypercone given by hardt and simon in @xcite in certain angular neighborhood of the cone under consideration . more flexible than homogeneous foliations are the isoparametric ones . in 1985 , ferus and karcher @xcite were able to construct the characterization foliation by homothetic minimizing hypersurfaces following an ode way for almost all " minimal isoparametric hypercones of ot - fkm type , and hence showed each of them minimizing . thereafter , to figure out a complete classification of minimizing isoparametric hypercones became important in the subject . based on the method of @xcite , an explicit list was given in @xcite . unfortunately , the arguments for strict minimality in his 9 were invalid due to the incorrect remark 9.17 on the correct formula in @xcite . we shall give an alternative proof in 2 . since a tangent cone ( at some point ) of an area - minimizing rectifiable current is itself area - minimizing ( theorem 5.4.3 in @xcite , also see theorem 35.1 and remark 34.6 ( 2 ) in @xcite ) , the study of minimizing cones of higher codimensions is of equal importance . besides minimal isoparametric hypersurfaces , another class of minimal submanifolds associated with isoparametric foliations are focal submanifolds . by the efforts of @xcite , cones over focal submanifolds of homogeneous isoparametric foliations with @xmath3 distinct principal curvatures and multiplicities @xmath4 had been shown minimizing . however , it seems more natural to consider cones over general focal submanifolds simultaneously and we shall do so in 3 . quite interestingly , for some of isoparametric foliations , their minimal isoparametric hypersurfaces can not generate minimizing cones , but the corresponding focal submanifolds can . + taking advantage of focal submanifolds of isoparametric foliations in spheres , we establish + * theorem 1 . * _ cones over focal submanifolds of isoparametric foliations in unit spheres with @xmath5 and @xmath4 are area - minimizing . _ lots of them are inhomogeneous v.s . those equivariant minimizing cones constructed in @xcite . it should be remarked that for some inhomogeneous isoparametric foliations their focal submanifolds can possibly be homogeneous . for details readers are referred to @xcite . for @xmath6 , both @xmath7 and @xmath8 , the cones over the focal submanifolds @xmath9 and @xmath10 , are minimizing and of dimension @xmath11 in @xmath12 . moreover , since the vanishing angle is less than @xmath13 @xmath14 , their union @xmath15 is minimizing as well . when @xmath16 for @xmath17 or @xmath18 is nonorientable , @xmath19 is minimizing in the sense of mod @xmath20 ( see @xcite ) . for @xmath5 and @xmath21 , @xmath9 is diffeomorphic to @xmath22 and @xmath10 is isometric to @xmath23 ( see @xcite ) . it can be proved without difficulty that the latter is orientable if and only if @xmath24 is even . our investigation can also work for @xmath25 , and leads to + * theorem 2 . * _ cones over focal submanifolds of isoparametric foliations with @xmath26 and @xmath4 are area - minimizing . _ + + * question . * it will be interesting to ask whether cones over focal submanifolds of isoparametric foliations with @xmath3 and @xmath27 are minimizing . to resovle the question , it seems necessary to introduce new ideas and tools . + for cones over products of focal submanifolds in spheres , the following can be proved + * theorem 3 . * _ cones over the minimal products of two focal submanifolds ( of isoparametric foliations ) with @xmath3 and @xmath4 are area - minimizing . _ + to include cases of @xmath3 and @xmath28 , we have + * theorem 4 . * _ cones , of dimension no less than @xmath29 , over the minimal products of two focal submanifolds corresponding to @xmath3 are area - minimizing . _ by a revisit to proofs of both theorems in [ added ] , we generalize theorem 3 and theorem 4 with the same conclusions for cones over the minimal products of multiple focal submanifolds . further , we remove the restriction on @xmath30 for the generalized theorem 4 in [ g2 ] . in fact one can also investigate cones over products of multiple minimal isoparametric hypersurfaces , and products among focal submanifolds and minimal isoparametric hypersurfaces . the paper is organized as follows . in 2 we review the classification of isoparametric foliations , ferus and karcher s result on isoparametric minimizing hypercones and lawlor s classification of homogeneous minimizing hypercones . as a result , classification of isoparametric minimizing hypercones and the property that isoparametric minimizing hypecones are all strictly area - minimizing easily follow . in 3.1 we briefly go through the ideas of lawlor s curvature criterion , then apply it to cones over focal submanifolds in 3.2 and 3.3 for @xmath5 , and @xmath31 , @xmath32 respectively . in 4 , we first consider cones over products of two focal submanfolds , next establish theorem 3 and theorem 4 , then extend both to the case of products of several focal submanifolds in [ added ] , and finally include the case involving @xmath33 in [ g2 ] . a closed ( embedded ) hypersurface @xmath34 in the unit sphere @xmath35 is called isoparametric , by e. cartan , if it has constant principal curvatures . moreover , if @xmath34 is isoparametric , then so are its parallel hypersurfaces . in this way , a foliation of hypersurfaces appears with two exceptional leaves @xmath18 focal submanifolds of higher codimensions . as it is well known , there is one and only one minimal hypersurface among the parallel isoparametric hypersurfaces in an isoparametric foliation . let @xmath36 be a unit normal vector field along @xmath34 , @xmath30 the number of distinct principal curvatures of @xmath34 , @xmath37 @xmath38 the principal curvatures with respect to @xmath36 and @xmath39 the multiplicity of @xmath40 . by a purely topological method , mnzner proved an elegant result that @xmath30 must be @xmath41 or @xmath32 and @xmath42 ( indices mod @xmath30 ) . according to values of @xmath30 , we have the following . + @xmath43 . the foliation is a trivial one , namely it is given by level sets of a height function restricted to @xmath44 with respect to a direction . regular leaves of the foliation are of @xmath45 type and the corresponding focal submanifolds are great spheres @xmath46 and @xmath47 . @xmath48 has to take values among @xmath49 and @xmath50 . cartan showed that the foliations are homogeneous ( see , for example , @xcite ) . moreover , all isoparametric hypersurfaces in spheres with 3 principal curvatures are precisely the tubes of constant radius over the standard veronese embedding of @xmath51 for @xmath52 , @xmath53 , @xmath54 ( quaternions ) , @xmath55 ( cayley numbers ) in @xmath56 , @xmath57 , @xmath58 , @xmath59 , respectively . @xmath60 . @xmath61 by mnzner @xcite and @xmath62 has to be @xmath63 or @xmath20 by abresch @xcite . for @xmath64 , dorfmeister and neher @xcite proved that the foliation is homogeneous . very recently , miyaoka @xcite was able to gain the same conclusion for @xmath65 . @xmath5 . by the recent beautiful results of cecil - chi - jensen @xcite and chi @xcite , the classification gets complete . such a foliation must be either of ot - fkm type or homogeneous with @xmath66 or @xmath67 . here the ot - fkm type means that leaves of the foliation are given by @xmath68\ ] ] where @xmath69 for self - adjoint endomorphisims @xmath70 , for @xmath71 , with relation @xmath72 . such structure exists only when @xmath73 and @xmath74 ( see @xcite ) . its multiplicities satisfy @xmath75 . + among foliations of ot - fkm , the first inhomogeneous foliation occurs when @xmath76 for @xmath77 . all other inhomogeneous foliations appear when @xmath78 and @xmath79 . theorem 2 of @xcite states that each inhomogeneous minimal isoparametric hypersurface @xmath80 of ot - fkm type leads to a minimizing hypercone @xmath81 in fact their argument works for all minimal isoparametric hypersurface of ot - fkm type with @xmath79 . + combining the classification theorem of isoparametric foliations mentioned above , theorem 2 of @xcite and lawlor s classification table on homogeneous minimizing hypercones verifies the classification given by @xcite . + lawlor s classification of homogeneous minimizing hypercones + * theorem . * ( @xcite ) let @xmath80 be a minimal isoparametric hypersurface in @xmath82 . then @xmath83 is minimizing if and only if @xmath84 and @xmath85 or @xmath86 . by 4.3 of @xcite , lawlor s criterion about @xmath83 is not only sufficient but also necessary for a minimal isoparametric hypersurface @xmath80 . moreover , whenever the criterion holds , the cone becomes strictly area - minimizing automatically . hence the next statement follows . + + * proposition . * ( @xcite ) let @xmath80 be a minimal isoparametric hypersurface in @xmath82 . if @xmath83 is minimizing , it is actually strictly area - minimizing . it is well known that focal submanifolds of isoparametric foliations are minimal submanifolds in spheres . they generate minimal cones . based on their special second fundamental forms , we apply lawlor s criterion to show theorem 1 . to be self - contained , we briefly review lawlor s method in 3.1 . then 3.2 is devoted to a proof of theorem 1 . along the same line , homogeneous cases for @xmath31 and @xmath32 will be discussed in 3.3 . let us recall some notations . let @xmath87 be a submanifold of dimension @xmath88 in @xmath82 and @xmath89 . fix @xmath90 . let @xmath91 be the @xmath92-dimensional great sphere which intersects @xmath87 at @xmath93 orthogonally . for @xmath94 , denote by @xmath95 the open @xmath96-disk centered at @xmath93 in @xmath91 . then @xmath97 is called the @xmath98 through @xmath93 . conventionally , we leave out the origin , so that one can talk about nonintersecting normal wedges " . we name @xmath99 the @xmath100 of @xmath101 . the @xmath102 of @xmath101 at @xmath93 means the largest angular radius @xmath96 for @xmath103 intersecting @xmath101 only in the ray @xmath104 . + @xmath96-normal wedges of a cone since the unique minimizing foliation for an area - minimizing hypercone in @xcite has the property of homothety , so do its perpendicular integral curves ( outside the origin ) . the latter naturally induces an area - nonincreasing projection to the cone . instead , lawlor looked for similar structure in some @xmath96-angular neighborhood of a minimal cone @xmath101 rather than in @xmath105 . if the boundary of the neighborhood happens to be mapped to the origin under the projection , then one can send everything outside the neighborhood to the origin . in this way an area - nonincreasing projection can be produced . lawlor considered the structure given by rotation of a suitable curve @xmath106 in each normal wedge @xmath103 for @xmath90 . positive homotheties of the surface rotated by @xmath106 are required to foliate the normal wedge . under the assumption that elements of @xmath107 do not intersect each other , the existence of his preferred area - nonincreasing projection in the @xmath96-angular neighborhood is equivalent to that the o.d . ine . : @xmath108 where @xmath109 for @xmath110 and @xmath111 means the second fundamental form for a unit normal @xmath112 to @xmath101 at @xmath93 , has a solution which can reach zero for each @xmath112 . letting @xmath112 range over all unit normals at @xmath93 , the requirement becomes that @xmath113 where @xmath114 supports a solution which reaches zero . in fact it can be shown that the solution @xmath115 which attains zero fastest , if exists , must satisfy the equality in . suppose @xmath115 gets zero at @xmath116 . we call @xmath117 the vanishing angle of @xmath101 . @xmath118 stands for the narrowest ( uniform ) size for the above construction for the preferred area - nonincreasing projection . in summary , there are two things to check for applying lawlor s criterion : \1 . there exists a finite vanishing angle @xmath118 . here we would like to remark that may support no solutions which can touch zero ; and \2 . the @xmath118-normal wedges do not intersect . + a technical point is how to control @xmath119 in practice . corollary 1.3.3 in @xcite says that , for @xmath120 and @xmath121 $ ] , @xmath122 where @xmath123 the inequality is sharp ; for certain matrices , equality holds for all @xmath124 . ( for example , for the classical coassociative lawson - osserman cone in @xmath125 , see @xcite . ) note that @xmath126 is nonincreasing in @xmath127 . limiting @xmath128 leads to corollary 1.3.4 @xmath129 based on and , lawlor considered @xmath130 @xmath131 respectively . let @xmath132 and @xmath133 be the corresponding vanishing angles . then @xmath134 lawlor gained the following table for @xmath135 and @xmath136 . + + here are some notes : @xmath137 , \ \ \ \label{3.7}\\ \text{but the vanishing angle based on the method does not exist in general . } \ \ \ \nonumber\\ \text{however , since the equality of } \eqref{str}\text { and } \eqref{nstr } \ \text{may not be attained , } \ \ \ \ \ \ \ \ \ \label{3.8}\\ \text{slots filled with `` * * * " could still support actual vanishing angles . } \ \ \ \ \ \ \ \ \ \ \ \nonumber\\ \text{\ , \ , \ \ \ \ values in parentheses are tentative , pending on more numerical analysis . } \label{3.9 } \end{aligned}\ ] ] generally , using gives more accurate data . nevertheless , has a key advantage , proposition 1.4.2 in @xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath24 be the dimension of a minimal cone under consideration , @xmath138 explained in , and @xmath139 the vanishing angle for depending on @xmath140 . then for any @xmath141 , @xmath142 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ one can make use of this property to estimate vanishing angles for @xmath143 . another useful property is the following monotonicity : @xmath144 which comes from the monotonicity of @xmath145 in @xmath138 . similarly , one can define @xmath146 and have @xmath147 now we are ready to proceed . let @xmath9 and @xmath10 be focal submanifolds of an isoparametric foliation of @xmath5 . main points are : + \1 . @xmath148-normal wedges of @xmath149 or @xmath8 are nonintersecting . + \2 . in the isoparametric setting , @xmath118 only depends on @xmath140 and we can check everything merely at one point . if vanishing angle @xmath118 exists and moreover is less than @xmath148 , then @xmath149 or @xmath8 is minimizing . + \3 . note that @xmath150 is strictly decreasing when @xmath151 . so , for @xmath152 , we have @xmath153 by , and the table of upper bounds of vanishing angles , it follows that @xmath154 + \4 . hence , @xmath155 for @xmath156 and @xmath157 . according to lawlor s table , @xmath158 for @xmath159 and @xmath157 thus , when @xmath160 and @xmath157 , @xmath161 + \5 . for our case , @xmath9 is of dimension @xmath162 , so @xmath163 a very delightful and special property of focal submanifolds is that , for a point @xmath93 of @xmath9 and any unit normal @xmath112 to @xmath149 at @xmath93 , the second fundamental form for @xmath112 with respect to certain orthonormal basis of @xmath164 is @xmath165 + hence @xmath166 therefore , when @xmath167 , by , and , lawlor s criterion applies and consequently @xmath168 are minimizing . + \6 . when @xmath169 , @xmath170 equals @xmath171 for @xmath149 and @xmath172 for @xmath8 . the vanishing angle exists for the latter and @xmath173 , whereas the former seems subtle because of encountering ( * * * ) " . using the fact @xmath174 instead of the control @xmath175 , one can figure out that vanishing angle for the former exists and is less than @xmath176 . thus , our proof of theorem 1 is complete.@xmath177 in both cases , @xmath61 . theorem 2 says precisely the following . + + * theorem 2@xmath178 . * for @xmath179 or @xmath180 , cones over focal submanifolds are minimizing . with respect to a unit normal vector at a point in a focal submanfold , the second fundamental form ( shape operator ) would be similar to . they are ( cf . @xcite ) @xmath181 and @xmath182 respectively . hence @xmath183 and @xmath184 . according to lawlor s table and , vanishing angle @xmath118 exists and @xmath185 for each of them . namely , the @xmath118-normal wedges over the focal submanifold under consideration are nonintersecting . so the statement stands . _ remark_. by cartan , for @xmath186 , @xmath187 is isometric to @xmath188 with constant gaussian curvature ; for @xmath189 , @xmath187 is diffeomorphic to @xmath190 ( see , for example , @xcite ) . given two focal submanifolds @xmath191 and @xmath192 for isoparametric foliations of spheres , with @xmath193 respectively . define an embedding @xmath194 by @xmath195 we write @xmath196 and @xmath197 short for @xmath198 and @xmath199 for convenience . together with @xmath200 , orthonormal bases @xmath201 and @xmath202 of @xmath203 and @xmath204 form an orthonormal basis @xmath205 of the normal space of @xmath206 at @xmath207 in @xmath208 . let @xmath209 be the symbol of shape operators . then @xmath210 , @xmath211 and @xmath212 have the following relations : @xmath213 and @xmath214 also note that there are three levi - civita connections @xmath215 and , for tangent vector fields @xmath216 and @xmath217 around @xmath218 , we have @xmath219 therefore , @xmath220 and trace@xmath221 . these imply that , @xmath222 minimally embeds @xmath34 into @xmath208 . at @xmath223 , let @xmath224 where @xmath225 is a unit normal to @xmath226 at @xmath196 in @xmath227 and @xmath228 a unit normal to @xmath229 at @xmath197 in @xmath230 , such that @xmath231 attains the smallest ( nonzero ) for @xmath232 let @xmath233 . there are three possibilities to consider for smallest @xmath234 . if none of these three cases support a finite @xmath234 , we define @xmath235 . suppose @xmath236 ( the actual situation for @xmath237 ) . then @xmath238 . * ( i). @xmath239 . recall that ( @xcite ) along any normal geodesic of @xmath226 in @xmath227 , it takes length @xmath240 to arrive at its pairing focal submanifold . so before @xmath241 it can not get back to @xmath226 . the same holds for @xmath229 in @xmath230 . it follows that @xmath242 since @xmath243 , it implies that @xmath244 can not belong to @xmath245 . therefore , @xmath246 so that @xmath247 . thus , @xmath248 which implies , by the relation @xmath249 , that @xmath250 now @xmath251 , \nonumber \end{aligned}\ ] ] which , together with , implies that @xmath252.\ ] ] by adding @xmath253 to both sides of , we have @xmath254\\ & = & 1.\nonumber \end{aligned}\ ] ] comparing with leads to @xmath255 to seek for the smallest @xmath234 , we consider the chance @xmath256 notice that @xmath257 . by ranges of @xmath258 and @xmath259 , @xmath260 _ _ r__emark . in fact , when @xmath261 , produces infinity . however , in this case , the realizable below asserts the finiteness of @xmath262 , ( ii). @xmath263 . ( similar for @xmath264 . ) for this case , we have @xmath265 as argued in ( i ) * , @xmath259 can not belong to @xmath245 and assume @xmath266 so that @xmath267 similarly , @xmath268 and @xmath269 again , for the smallest @xmath234 , we focus on the likelihood @xmath270 by the range of @xmath258 , @xmath271 * ( iii). @xmath272 . in this case , @xmath273 which implies @xmath274 it is not hard to see that , only when @xmath275 , there is possibly a finite solution @xmath276 and @xmath277 hence , for @xmath278 , @xmath279 is no less than , by definition , the smallest quantity in , and . let @xmath280 be unit normals to @xmath226 and @xmath229 at @xmath196 and @xmath197 in @xmath227 and @xmath230 , and @xmath281 , @xmath282 the corresponding shape operators respectively . set @xmath283 then , for a unit normal @xmath284 with @xmath285 , the shape operator @xmath286 of @xmath34 at @xmath287 in @xmath208 is @xmath288 & o\\ o & \frac{1}{\mu } \big[c_0\lambda i_{k_2}+c_2 a_{\xi_2}\big ] \end{pmatrix}.\ ] ] since @xmath226 and @xmath229 are minimal , trace@xmath289 and trace@xmath290 . set @xmath291 . it follows that @xmath292 we make the following observations from , and : @xmath293 thus , by the expression , we always have @xmath294 for @xmath295 or @xmath29 and @xmath296 , vanishing angle @xmath297 exists according to lawlor s table and the fact that @xmath298 when @xmath299 , we seek for the existence of vanishing angle @xmath300 . note that , when the dimension of a cone equals @xmath301 , lawlor s table confirms the existence of @xmath300 for @xmath302 . recall : for @xmath141 @xmath303 since @xmath304 is strictly increasing for @xmath305 , one can find that , when @xmath306 , @xmath307 take @xmath308 and fix @xmath309 . then @xmath310 in the latter term of exists , and so does @xmath311 . by monotonicity , if @xmath312 , @xmath313 exists . in particular , according to , the existence of @xmath314 for @xmath315 is guaranteed . as a result , @xmath316 exists for the case of @xmath299 . hence , @xmath317 and @xmath118 always exist for @xmath318 . if we could show that @xmath321 where the normal radius @xmath319 is given at the end of 4.1 , then lawlor s criterion applies . by appendix , we have @xmath322 assume that @xmath323 from now on . one can see that @xmath324 hence , @xmath325 and therefore @xmath326 [ rknice ] note that @xmath124 in [ nr ] is allowed to be negative . it implies that the upper bound @xmath327 is in fact uniform for @xmath328 where @xmath329 corresponds to any intersection point ( if existed , except @xmath330 ) of @xmath331 and a great circle perpendicular to @xmath331 through @xmath218 . now we are at a position to prove theorem 3 . + proof of theorem 3 . * note that @xmath332 in the case . since @xmath333 exists for @xmath318 , we have , by , , and , that @xmath334 then @xmath335 since @xmath336 we have@xmath337 and thus complete the proof . @xmath177 + let us assume @xmath338 and figure out when @xmath337 holds . note that @xmath339<\tan 9.55^\circ<0.1683 .\ ] ] combined with , and , the existence of @xmath340 supports that @xmath341 < \dfrac{12}{s+1}\tan\left [ \theta_c\big(12,\ \sqrt{\dfrac{44}{3 } } \big)\right].\ ] ] then @xmath342 hence , with the agreement that @xmath323 , it is sufficient to show @xmath343 which is equivalent to @xmath344 ^ 2 - \left ( s-\frac{k_1}{2 } \right)^2 \left [ 24(s+1)0.1683 \right]^2 > 0.\ ] ] + proof of theorem 4 . * when @xmath345 , a proof is given in that of theorem 3 . consider now that @xmath346 and @xmath347 . to determine the sign of polynomial , we run the following codes in mathematica : + and it can be seen that , whenever @xmath299 for , we have hold . let us check the cases of @xmath348 for @xmath346 ( which can imply the cases for @xmath349 ) . by lawlor s table , @xmath350 and @xmath351 . then direct computations show that @xmath352 and @xmath353 respectively . so , @xmath354 and the proof gets complete . @xmath177 + we exhibit the ideas about generalizations of theorem 3 and theorem 4 only for products over three focal submanifolds . the case involving more components can be similarly done by induction . given three focal submanifolds @xmath191 , @xmath192 and @xmath355 for isoparametric foliations of spheres . define @xmath356 by @xmath357 @xmath358 then @xmath359 is a minimal embedding , which follows from that @xmath360 where map @xmath222 is given for two minimal submanifolds at the beginning of this section . let @xmath361 . with the above understanding , it easily follows , from expressions and , that the norm square of shape operator of @xmath362 at every point for any unit normal is no more than @xmath363 . hence , vanishing angles @xmath364 and @xmath365 corresponding to that in [ eva ] and those in proofs of theorem 3 and theorem 4 exist . without loss of generality , assume that @xmath366 . let @xmath319 and @xmath367 be the normal radii of @xmath368 in @xmath208 and @xmath362 in @xmath369 respectively . we stick to the notation @xmath370 . set @xmath371 , @xmath372 and @xmath373 , . then , applying the discussions of * ( i ) * , * ( ii ) * and * ( iii ) * to @xmath374 , we have , according to appendix and remark [ rknice ] , that @xmath375 hence , we gain an inheritable relation ( with the property in remark [ rknice ] ) @xmath376 based on the proofs of theorem 3 and theorem 4 , it then follows correspondingly that @xmath377 the descendent and coupled property in remark [ rknice ] are crucial for this procedure to be valid for the induction on the number of focal submanifolds . due to the distinct behaviors of corresponding focal submanifolds , we save the discussions about @xmath33 separately in this subsection . since the case of products over spheres has been classified by @xcite . we only consider the mixed type . more explicitly , let @xmath378 be focal submanifolds in spheres corresponding to @xmath379 , and @xmath380 focal submanifolds for @xmath33 . then @xmath381 with @xmath382 is what we are concerned about . * * @xmath383 . namely @xmath384 and @xmath385 . denote the normal radius of the minimal submanifold @xmath34 in the sphere by @xmath386 . assume that @xmath387 and set @xmath388 . then * ( i ) * , * ( ii ) * and * ( iii ) * give a lower bound @xmath389 of @xmath390 . although the last term of the following has no apparent geometric meaning in the current case , one can deduce from remark [ stronger ] that @xmath391 therefore , @xmath392 * ( b ) . * @xmath393 with @xmath384 and @xmath394 . we use symbols @xmath395 and @xmath386 to represent the normal radii of minimal products @xmath396 and @xmath34 in the corresponding spheres , respectively . assume that @xmath397 and set @xmath398 . then , according to 5.1 of @xcite , @xmath399 . also note that the minimal @xmath396 is symmetric about the origin . hence , @xmath400 has the uniform upper bound property mentioned in remark [ rknice ] . therefore , @xmath401 observe that by induction relation can lead to the same result as for @xmath34 . thus , by comparing and with , and , we can include @xmath33 in the generalized theorem 4 with no restriction on @xmath402 , under the same dimension assumption . ( results of the products of purely spheres are due to @xcite . ) as a conclusion remark , we point out that , if @xmath403 with @xmath404 , the condition @xmath345 is sufficient for @xmath405 to be minimizing in the corresponding euclidean space . in [ nr ] we figure out four quantity candidates to be a lower bound of @xmath262 . to make arguments in [ cp ] , [ added ] and [ g2 ] simple , a proof of the following useful inequality is given here . [ stronger ] for integers @xmath414 and @xmath415 , the above inequality can be improved to @xmath416 here we think of a nonzero number divided by zero as infinity . more precisely , when @xmath417 , @xmath418 with equality if and only if @xmath419 . * acknowledgments*. we thank professor gary lawlor for permission of using his graph and tables from @xcite , professor michael kerckhove and doctor chao qian for useful comments . the second author wishes to thank professor leon simon sincerely for many helpful explanations about @xcite . this work was sponsored in part by nsfc ( no . 11331002 , 11526048 , 11601071 ) , the fundamental research funds for the central universities , and the srf for rocs , sem .	this paper is concerned with cones associated with isoparametric foliations in unit spheres . in particular , infinitely many new minimizing cones were constructed . they are cones over focal submanifolds of isoparametric foliations . a major portion of them are inhomogeneous . further , we show that suitable products of multiple focal submanifolds can generate minimizing cones as well . [ 0em ] [ 1.5em ]
we are interested in an active imager , such as a laser radar , comprised of a transmitter and a receiver in which we control the object illumination and form the image with the receiver . for such systems , the spatial resolution is a function of two antenna patterns , viz . , the transmitter s illumination pattern on the object _ and the receiver antenna pattern , set by the diffraction limit of its optics , projected onto the object . when floodlight illumination is employed , so that the entire object is bathed in light , the resolution limit is set by the receiver s rayleigh diffraction bound . if the transmitter and receiver are co - located , they can share the same optics so that their antenna patterns have identical rayleigh bounds whose product gives the overall resolution behavior . alternatively , if a small transmitter is located much closer to the object than is the receiver , it is possible to project very small spots onto the object to be imaged . if this is done in a precision scan , so that the receiver knows exactly where the transmitter is pointing at any instant in time , a simple energy measurement at the receiver will realize resolution limited by the transmitter s antenna pattern , regardless of the receiver s own rayleigh bound . however , creating that precision scan , and relaying the scan positions to the receiver , could easily be a major challenge , especially if the transmitter is mounted on a small unmanned air vehicle . the postselection technique of giovannetti _ et al.circumvents that problem by allowing the scan pattern to be arbitrary , even random , so long as the object s region of interest is covered . _ _ suppose that a focused transmitter emits a @xmath4-going , quasimonochromatic , paraxial , linearly - polarized laser pulse with scalar complex envelope @xmath5 for @xmath6 , where @xmath7 is the transverse coordinate vector , @xmath8 is the transmitter aim angle , and @xmath9 is the wave number at the center wavelength @xmath10 . we will normalize @xmath11 to have units @xmath12 , and take the pulse shape @xmath13 to satisfy @xmath14 , so that @xmath15 is the average transmitted photon number @xcite . this pulse transilluminates an object @xmath16-m away from the transmitter @xcite . the light that passes through the object is then collected by a diffraction - limited circular lens of diameter @xmath17 located @xmath18-m in front of the object . the focal length of this lens is such that it casts an image of the object at a distance @xmath19 beyond the lens . neglecting the propagation delay and correcting for image inversion , the photon - flux density in the image plane is then @xmath20 where @xmath21 is the object s field - transmission function , @xmath22 is the image magnification @xcite , and we see the transmitter and receiver s circular - pupil antenna patterns . to exhibit the sub - rayleigh resolution capability of the scheme from @xcite we shall assume that : ( 1 ) the transmitter s antenna pattern fully resolves all significant features in @xmath21 @xcite ; ( 2 ) the image - plane photon counting array has pixels of area @xmath23 sufficiently small that diffraction , rather than pixel size , limits image resolution ; and ( 3 ) the photon counting array outputs pixel counts taken over the full @xmath3-s - long extent of @xmath13 . for a given illumination angle @xmath24 , the pixel counts are then statistically independent , poisson random variables with mean value , for the pixel centered at @xmath25 , given by @xmath26 where @xmath27 is the detector s quantum efficiency . if photon counts are collected from the pixel at @xmath25 while @xmath24 is randomly scanned over the object region , the unconditional probability of getting @xmath0 counts from that pixel is @xmath28 where @xmath29 is the scan pattern s probability density function . postselecting those pixels for which @xmath0 counts have been registered , we get an image @xmath30 . for @xmath31 , the poisson distribution is monotonically decreasing with increasing @xmath0 , whence @xmath32 where we have suppressed multiplicative constants and ignored the exponential term as it is independent of @xmath0 . here we see that the postselected image contains @xmath33 convolved with the @xmath0th power of the receiver s airy disk pattern , i.e. , a point - spread function that is @xmath1@xmath2 narrower than the airy disk itself when compared on a main - lobe area basis @xcite . note that violating the preceding monotonicity condition can lead to the `` donut - hole '' effect exhibited in the experiments described below . we demonstrate the concept of sub - rayleigh imaging with the setup shown in fig . [ f : setup](a ) . the object to be imaged in transmission was part of a u.s . air force ( usaf ) resolution target consisting of alternate opaque and clear stripes of width 125@xmath34 m ( 4 line pairs / mm ) , as indicated by the arrow in fig . [ f : image](a ) . a 532-nm laser was mounted on an @xmath35 translation stage that provided scan coverage over the entire object with a 20-@xmath34m - radius focused spot . we imaged the object through a @xmath36 = 25-cm diffraction - limited lens set in a 2-mm - diameter aperture . the optics provided 5.3@xmath37 image magnification , yielding a 660-@xmath34m - wide strip at the image plane . under conventional ( entire - object ) illumination , shown in the setup of fig . [ f : setup](b ) , the rayleigh diffraction bound for the imaging system at the image plane was 1.86 mm , which is @xmath12.8@xmath37 larger than the stripe width . figure [ f : image](b ) shows the conventional - illumination image that was obtained using standard photodetection ( with all events counted ) : the stripes are unresolved , as expected . note that with full illumination of the object , we were not able to go beyond the rayleigh bound even with the @xmath0-photon detection scheme , indicating that focused illumination is a necessary requirement . the detector was a compact @xmath38 si single - photon avalanche diode ( spad ) array fabricated with a complementary metal oxide semiconductor ( cmos ) process @xcite . the si spad was a @xmath39-@xmath40 junction reverse - biased above its breakdown voltage and operated in the geiger mode with a detection efficiency of @xmath130% at 532 nm . each pixel of the cmos spad array consisted of one spad with its front - end active quenching and resetting electronics and a digital counting circuitry for in - pixel pre - processing . the pixel pitch for the array was 100@xmath34 m , and the spad had a fill factor of only 3.1% at each pixel due to the presence of on - chip electronics . owing to the large separation between spads we did not observe any cross talk . the average pixel dead time ( including all the electronic circuitry ) was 300ns . each spad delivered a digital output pulse for every single - photon detection event with no readout noise . the in - pixel counting circuitry would compute the number of single - photon events within its user - selectable integration time of 1@xmath34s or more and store the tally in an in - pixel memory cell . by measuring incident photons over a long integration time , @xmath0-photon sensitivity in the time domain can be achieved at the single - pixel level . the array readout was performed through an 8-bit data bus without interrupting the next 1024-pixel frame of photon - counting integration , and the maximum frame rate was @xmath41/s . a typical integration time of each frame was tens of @xmath34s . to implement sub - rayleigh imaging in fig . [ f : setup](a ) we manually scanned the focused beam in a random pattern , making sure that there was coverage for the entire area of interest . at each scan location , we recorded over 8000 measurement frames for image averaging that took less than 1 second to accomplish . the incident power was adjusted to have an average peak photocount @xmath42 = 14 per integration time ( for one pixel ) . for each measurement frame , each pixel with exactly @xmath0 photocounts ( after dark - count subtraction , measured separately ) was tagged as having an @xmath0-photon event . all other pixels were then tagged for zero @xmath0-photon events . the measurement process was then repeated at a different scan location until the object of interest was fully scanned . figure [ f : image](c ) shows the resultant image for @xmath0 = 23 , revealing the three stripes that were lost under conventional illumination in fig . [ f : image](b ) . the color scale of fig . [ f : image](c ) has a large range to accommodate several pixels with very high event counts and therefore image details ( with lower event frequencies ) are obscured . figure [ f : image](d ) shows the 3-d intensity profile of the same @xmath43 image of fig . [ f : image](c ) , except that we cap the event occurrence at 800 for pixels with over 800 events in order to make the lower - count pixels more visible , thus revealing the three stripes very clearly . according to theory , the expected enhancement of @xmath2 should yield a sub - rayleigh resolution of @xmath11.86/@xmath44 = 0.4 mm is qualitatively borne out by our results . note that the sub - rayleigh resolution still exceeds the 106-@xmath34m - limit set ( after magnification ) by the focused illumination at the object , as expected . we chose @xmath43 to be substantially larger than @xmath42 = 14 to avoid the donut - hole " problem . to illustrate this issue , fig . [ f : pointsource ] shows images of a point source obtained under various measurement conditions . the aperture diameter of the imaging optics in fig . [ f : setup](a ) was set to 3 mm with the same overall image magnification of 5.3 so that the rayleigh bound at the image plane ( spad array ) was 1.2 mm . we removed the usaf resolution target so that the 20-@xmath34m - radius focused spot at the object plane served as the point source , and no scanning was necessary for these images . we took @xmath132,000 measurement frames , recorded the photocounts at each pixel for each frame , binned them accordingly after subtracting dark counts , and processed the data . figure [ f : pointsource](a ) shows the cross section of the rayleigh - bound image of the point source through the 3-mm - diameter aperture obtained by including all photocounts to yield an intensity profile averaged over the @xmath132000 frames . we measured @xmath45 and , as an indicator for the image size , we obtained the full - width at half - maximum ( fwhm ) of @xmath11 mm . figures [ f : pointsource](b)(d ) are cross - sectional profiles obtained by selecting exactly @xmath0-photocounts for @xmath46 9 , 15 , and 23 , that are respectively smaller than , equal to , and greater than @xmath42 . for @xmath47 in ( b ) , the center portion of the point - source image usually received more than the threshold level @xmath0 and therefore had few exactly @xmath0-photon events . on the other hand , the photocounts away from the center decrease from @xmath42 until the photocount average matches the threshold @xmath0 where it shows a peak , and hence the image has a donut - hole " shape @xcite . for @xmath48 in ( c ) , the profile is single - peaked and looks slightly narrower and steeper than the rayleigh - bound image in fig . [ f : pointsource](a ) which is much larger than the magnified size of the focused beam of 125 @xmath34 m ( fwhm ) at the image plane . when @xmath49 , as required for sub - rayleigh imaging , we observe in ( d ) that the profile for @xmath0 = 23 shows a much narrower peak with a fwhm width of @xmath10.4 mm that is smaller than the 1-mm width ( fwhm ) of the rayleigh - bound point source image in fig . [ f : pointsource](a ) . we also note that the @xmath0 = 23 event frequency is much lower than for @xmath0 less than or equal to @xmath42 , because such a large @xmath0 did not happen very often . our sub - rayleigh imaging has characteristics that are similar to @xmath0-photon interferometry , in which sub - wavelength ( @xmath50 ) interference patterns were obtained with coherent - state input and postselective @xmath0-photon detection @xcite . our point - source images suggest the following physical origin for the formation of the rayleigh diffraction bound and the basis for our sub - rayleigh imaging technique . the donut holes that form for @xmath47 are responsible for the diffractive spread of the imaging system . by removing the low-@xmath0 photocount events and those comparable to @xmath42 , one captures the much narrower profile for @xmath49 at the expense of longer acquisition times due to less frequent occurrences for high values of @xmath0 . we should also achieve similar results if we relax the measurement requirement from exactly @xmath0 for @xmath49 to a sum of all @xmath49 , whose resolution should be dominated by its lowest @xmath0 term due to its larger sub - rayleigh width ( @xmath51 ) and higher rate of occurrence . in conclusion , we have demonstrated sub - rayleigh imaging resolution using a classical light source , tight focusing on the object , and @xmath0-photon photodetection . the sub - rayleigh technique removes the low-@xmath0 image components that contribute to the diffractive spread of the imaging system . the measured resolution enhancement of @xmath1@xmath2 is in good agreement with theory . the sub - rayleigh technique may find applications in which active illumination can be implemented to obtain a higher resolution image at the expense of longer acquisition times . we thank j. le gout for help with the experiment . f. g. acknowledges the roberto rocca doctoral fellowship . this work was supported in part by the w. m. keck foundation center for extreme quantum information theory , the u.s . army research office under a multidisciplinary university research initiative grant , and the darpa quantum sensors program . v. giovannetti , s. lloyd , and l. maccone , science * 306 , * 1330 ( 2004 ) ; phys . * 96 , * 010401 ( 2006 ) ; s. l. braunstein , nature * 440 , * 617 ( 2006 ) . see , e.g. , m. i. kolobov , rev . * 71 , * 1539 ( 1999 ) ; m. i. kolobov ( ed . ) _ quantum imaging _ , ( springer , new york , 2006 ) . y. shih , ieee j. sel . topics in quantum electron . * 13 , * 1016 ( 2007 ) . lugiato _ et al . b * 4 , * s176 ( 2002 ) . a. n. boto , _ et al . * 85 , * 2733 ( 2000 ) . m. tsang , phys . * 102 , * 253601 ( 2009 ) . k. l. pregnell and d. t. pegg , j. mod . * 51 , * 1613 ( 2004 ) . p. r. hemmer , _ et al . * 96 , * 163603 ( 2006 ) ; a. muthukrishnan , m. o. scully , and m. s. zubairy , j. opt . b : quantum semiclass . opt . * 6 , * s575 ( 2004 ) . c. thiel * 99 , * 133603 ( 2007 ) . s. j. bentley and r. w. boyd , opt * 12 , * 5735 ( 2004 ) . et al . _ , lett . a * 366 , * 569 ( 2007 ) . k. wang and d .- z . cao , phys . a * 70 , * 041801(r ) ( 2004 ) . d. korobkin and e. yablonovitch , opt * 41 , * 1729 ( 2002 ) . e. yablonovitch and r. b. vrijen , opt * 38 , * 334 ( 1999 ) . a. peer _ et al . _ , opt . express * 12 , * 6600 ( 2004 ) . b. i. erkmen and j. h. shapiro , phys . rev . a * 77 , * 043809 ( 2008 ) . v. giovannetti , s. lloyd , l. maccone , j. h. shapiro , phys . a * 79 , * 013827 ( 2009 ) . without loss of generality , we use semiclassical photodetection theory in what follows , hence a classical field description suffices , see , e.g. , j. h. shapiro , ieee j. sel . topics in quantum electron . * 15 , * 1547 ( 2009 ) . standoff sensing is ordinarily done in reflection , but we assume transillumination here to match our experiment . laser radar operation would employ @xmath52 , hence image minification . we have specified @xmath19 to match our experiment . if @xmath53 is the minimum size of any significant @xmath21 feature , then we require @xmath54 . f. guerrieri , s. tisa , a. tosi , and f. zappa , proc . spie * 7536 , * 753605 ( 2010 ) . a. j. pearlman , _ et al . _ , opt . express * 18 , * 6033 ( 2010 ) . g. khoury , h. s. eisenberg , e. j. s. fonseca , and d. bouwmeester , phys . * 96 , * 203601 ( 2006 ) .	the rayleigh diffraction bound sets the minimum separation for two point objects to be distinguishable in a conventional imaging system . we demonstrate resolution enhancement beyond the rayleigh bound using random scanning of a highly - focused beam and @xmath0-photon photodetection implemented with a single - photon avalanche detector array . experimental results show resolution improvement by a factor @xmath1@xmath2 beyond the rayleigh bound , in good agreement with theory . the response of a diffraction - limited imaging system to a point - like source its point spread function ( psf)has extent inversely proportional to the entrance pupil s area . the image is obtained by convolving the system psf with the light distribution of the object . therefore , details finer than the psf s extent are lost under conventional ( entire - object ) illumination . the rayleigh diffraction bound sets the minimum separation for two object points to be distinguishable in the image . two classes of quantum strategies have been suggested to circumvent this bound . the first relies on techniques from quantum metrology @xcite , in which the image information is encoded into suitably tailored non - classical light beams @xcite . for example , quantum lithography @xcite exploits the effective de broglie wave - function of @xmath0 photons in a delicately crafted state to obtain an increase in resolution proportional to @xmath0 ( see @xcite for recent developments ) . these methods are typically extremely sensitive to photon loss and noise , because they rely on delicate quantum effects ( such as squeezing and entanglement ) . thus they are best suited to short - distance applications , such as microscopy , where losses can be controlled , as opposed to standoff sensing , such as laser radar operation over km or longer path lengths , for which substantial diffraction and atmospheric losses will be present . the second class of quantum strategies for beating the rayleigh diffraction bound exploit postselection @xcite to extract the high - resolution image associated with a non - classical component from classical - state light containing information about the object to be imaged @xcite . because postselection involves discarding part of the measurement data , these procedures inherently suffer detection inefficiency that increases the time required to acquire an image . however , their spatial resolution can nonetheless exceed the rayleigh diffraction bound . furthermore , because they employ classical - state ( laser ) light , these techniques degrade gracefully with increasing loss and noise , making them suitable for standoff sensing . in this letter we report the first experimental demonstration of one such technique , viz . , that of giovannetti _ et al . _ @xcite , in which the object is illuminated by a focused light source and scanned . the scanning pattern is irrelevant so long as the area of interest is covered , i.e. , a random scan pattern suffices . the image is formed using only those pixels that count exactly @xmath0 photons within a measurement time @xmath3 . the expected resolution improvement is @xmath2 over standard entire - object illumination , until the limit set by the focused - beam illumination . we begin with the theory for this technique .
average distance is one of the most important measurements characterizing complex networks , which is a subject attracting a lot of interest in the recent physics literature @xcite . extensive empirical studies showed that many , perhaps most , real networks exhibit remarkable small - world phenomenon @xcite , with their average distance grows as a function of network order ( i.e. , number of nodes in a network ) , or slowly @xcite . as a fundamental topological property , average distance is closely related to other structural characteristics , such as degree distribution @xcite , centrality @xcite , fractality @xcite , symmetry @xcite , and so forth . all these features together play significant roles in characterizing and understanding the complexity of networks . moreover , average distance is relevant to various dynamical processes occurring on complex networks , including epidemic spreading @xcite , target search @xcite , synchronization @xcite , random walks @xcite , and many more . in addition to the small - world behavior , other two prominent properties that seem to be common to real networks , especially biological and social networks , are scale - free feature @xcite and modular structure @xcite . the former implies that the networks obey a power - law degree distribution as @xmath0 with @xmath1 , while the latter means that the networks can be divided into groups ( modules ) , within which nodes are more tightly connected with each other than with nodes outside . in order to describe simultaneously the two striking properties , ravasz and barabsi ( rb ) presented a famous model @xcite , mimicking scale - free modular networks . many topological properties of and dynamical processes on the rb model have been investigated in much detail , including degree distribution @xcite , clustering coefficient @xcite , betweenness centrality distribution @xcite , community structure @xcite , random walks @xcite , among others . particularly , by mapping the networks onto a potts model in one - dimensional lattices , noh proved that the rb model is small - world @xcite . in this paper , we study the average distance in the rb model by using an alternative approach very different from the previous one @xcite . our computation method is based on the particular deterministic construction of the rb model . concretely , making use of the self - similar structure of the scale - free modular networks , we establish some recursion relations , from which we further derive the exactly analytical solution to the average distance . our obtained rigorous expression is compatible with the previous formula . we show that the rb model is small - world . we also show that the small - world behavior is a natural result of the scale - free and modular architecture of the networks under consideration . we first introduce the rb model for the scale - free modular networks , which are built in an iterative way @xcite . let @xmath2 stand for the network model after @xmath3 ( @xmath4 ) iterations ( i.e. , number of generations ) . initially ( @xmath5 ) , the model is composed @xmath6 ( @xmath7 ) nodes linked by @xmath8 edges forming a complete graph , among which a node ( e.g. , the central node in figure [ network ] ) is called hub ( or root ) node , and the other @xmath9 nodes are named peripheral nodes . at the second generation ( @xmath10 ) , @xmath9 replicas of @xmath11 are created with the @xmath9 peripheral nodes of each copy being connected to the root of the original @xmath11 . in this way , we obtained @xmath12 , the hub and peripheral nodes of which are the hub of the original @xmath11 and the @xmath13 peripheral nodes in the @xmath9 duplicates of @xmath11 , respectively . suppose one has @xmath14 , the next generation network @xmath2 can be obtained by adding @xmath15 copies of @xmath14 to the primal @xmath14 , with all peripheral nodes of the replicas being linked to the hub of the original @xmath14 unit . the hub of the original @xmath14 and the peripheral nodes of the @xmath9 copies of @xmath14 form the hub node and peripheral nodes of @xmath2 , respectively . repeating indefinitely the two steps of replication and connection , one obtains the scale - free modular networks . figure [ network ] illustrates a network @xmath16 for the particular case of @xmath17 . for the case of @xmath17 . the filled squares and circles represent the hub node and peripheral nodes , respectively . ] many interesting quantities of the model can be determined explicitly @xcite . in @xmath2 , the network order , denoted by @xmath18 is @xmath19 ; the degree @xmath20 $ ] of the hub node is the largest among all nodes ; the number of peripheral nodes , forming a set @xmath21 , is @xmath22 ; and the average degree is approximately equal to a constant @xmath23 in the limit of infinite @xmath3 , showing that the networks are sparse . the model under consideration is in fact an extension of the one proposed in @xcite and studied in much detail in @xcite . it presents some typical features observed in a variety of real - world systems @xcite . its degree distribution follows a power - law scaling @xmath0 with a general exponent @xmath24 belonging to the interval @xmath25 . its average clustering coefficient tends to a large constant dependent on @xmath6 ; and its average distance grows logarithmically with the network order , both of which show that the model is small - world . in addition , the betweenness distribution @xmath26 of nodes also obeys the power - law behavior @xmath27 with the exponent regardless of the parameter @xmath6 . particularly , the whole class of the networks shows a remarkable modular structure . these peculiar structural properties make the networks unique within the category of complex networks . as shown in the introduction section , average distance is closely related to many topological properties of and various dynamical processes on complex networks . in what follows , we will derive analytically the average distance of the scale - free modular networks by applying an alternative method completely different from that in @xcite . we represent all the shortest path lengths of network @xmath2 as a matrix in which the entry @xmath28 is the distance between nodes @xmath29 and @xmath30 that is the length of a shortest path joining @xmath29 and @xmath30 . a measure of the typical separation between two nodes in @xmath2 is given by the average distance @xmath31 defined as the mean of distances over all pairs of nodes : @xmath32 where @xmath33 denotes the sum of the distances between two nodes over all couples . notice that in eq . ( [ total01 ] ) , for a pair of nodes @xmath29 and @xmath30 ( @xmath34 ) , we only count @xmath28 or @xmath35 , not both . ( color online ) schematic illustration of the means of construction of the scale - free modular networks . @xmath36 is obtained by joining @xmath6 replicas of @xmath2 denoted as @xmath37 @xmath38 , which are connected to one another by linking all the peripheral nodes of @xmath39 ( @xmath40 ) to the hub node ( denoted by @xmath41 ) of @xmath42 . ] we continue by exhibiting the procedure of determining the total distance and present the recurrence formula , which allows us to obtain @xmath43 of the @xmath44 generation from @xmath45 of the @xmath3 generation . the studied network @xmath2 has a self - similar structure that allows one to calculate @xmath45 analytically . by construction ( see figure [ labeling ] ) , network @xmath36 is obtained by joining @xmath6 copies of @xmath2 that are labeled as @xmath42 , @xmath46 , @xmath47 , @xmath48 . using this self - similar property , the total distance @xmath43 satisfies the recursion relation @xmath49 where @xmath50 is the sum over all shortest path length whose endpoints are not in the same @xmath37 branch . the paths that contribute to @xmath50 must all go through the hub node @xmath41 , where the @xmath6 copies of @xmath2 are connected . hence , to determine @xmath51 , all that is left is to calculate @xmath50 . the analytic expression for @xmath50 , referred to as the crossing path length @xcite , can be derived as below . let @xmath52 be the sum of the lengths of all shortest paths whose endpoints are in @xmath53 and @xmath54 , respectively . according to whether the two branches are one link long or two links long , we split the crossing paths @xmath52 into two categories : the first category composes of crossing paths @xmath55 ( @xmath56 ) , while the second category consists of crossing paths @xmath57 with @xmath58 , @xmath59 , and @xmath60 . it is easy to see that the numbers of the two categories of crossing paths are @xmath9 and @xmath61 , respectively . moreover , any two crossing paths in the same category have the same length . thus , the total sum @xmath50 is given by @xmath62 having @xmath50 in terms of the quantities of @xmath63 and @xmath64 , the next step is to explicitly determine the two quantities . to calculate the crossing distance @xmath63 and @xmath64 , we give the following notation . for an arbitrary node @xmath65 in network @xmath2 , let @xmath66 be the smallest value of the shortest path length from @xmath65 to any of the @xmath67 peripheral nodes belonging to @xmath21 , and the sum of @xmath66 for all nodes in @xmath2 is denoted by @xmath68 . analogously , in @xmath2 let @xmath69 denote the distance from a node @xmath65 to the hub node @xmath41 , and let @xmath70 stand for the total distance between all nodes in @xmath2 and the hub node @xmath41 in @xmath2 , including @xmath41 itself . by definition , @xmath71 can be given by the sum @xmath72+(m-1)\,\sum_{v \in h_{g}}f_v(g)\nonumber \\ & = & ( m-1)\,f_{g}+n_g+m_g\,,\end{aligned}\ ] ] and @xmath73 can be written recursively as @xmath74\nonumber \\ & = & m_g+(m-1)(f_{g}+n_g)\,.\end{aligned}\ ] ] using @xmath75 , and considering @xmath76 and @xmath77 , the simultaneous equations ( [ bottom01 ] ) and ( [ hub01 ] ) can be solved inductively to obtain : @xmath78\ ] ] and @xmath79 with above obtained results , we can determine @xmath63 and @xmath64 , which can be expressed in terms of these explicitly determined quantities . by definition , @xmath63 is given by the sum @xmath80\nonumber \\ & = & \sum_{v \in h_{g}^{(2 ) } } \sum_{u \in h_{g}^{(1 ) } } h_u(g)+\sum_{u \in h_{g}^{(1)}}\sum_{v \in h_{g}^{(2)}}[1+f_v(g ) ] \nonumber \\&=&n_g\,m_g+(n_g)^2+n_g\,f_g\,.\end{aligned}\ ] ] inserting eqs . ( [ bottom02 ] ) and ( [ hub02 ] ) into ( [ cross03 ] ) , we have @xmath81\,.\ ] ] proceeding similarly , @xmath82\nonumber \\&=&2\,m^{2g-4}\left[m^2+(2g-3)m-2g+4\right]\,.\end{aligned}\ ] ] substituting eqs . ( [ cross04 ] ) and ( [ cross05 ] ) into ( [ cross01 ] ) , we get @xmath83 substituting eq . ( [ cross06 ] ) into ( [ total01 ] ) and using the initial value @xmath84 , we can obtain the exact expression for the total distance @xmath85 \,.\ ] ] the expression provided by eq . ( [ total03 ] ) is consistent with the result previously obtained @xcite . then the analytic expression for average distance can be obtained as @xmath86 average distance @xmath31 versus network order @xmath18 on a semi - logarithmic scale . the solid lines are guides to the eye . ] we have also checked our rigorous result provided by eq . ( [ apl02 ] ) against numerical calculations for different @xmath6 and various @xmath3 . in all the cases we obtain a complete agreement between our theoretical formula and the results of numerical investigation , see figure [ avedis ] . we continue to express the average distance @xmath87 as a function of network order @xmath18 , in order to obtain the scaling between these two quantities . recalling that @xmath75 , we have @xmath88 . hence eq . ( [ apl02 ] ) can be rewritten as @xmath89 in the infinite network order limit , i.e. , @xmath90 @xmath91 thus , for large networks , the leading behavior of average distance grows logarithmically with increasing network order . the above observed small - world phenomenon that the leading behavior of average distance is a logarithmic function of network order can be accounted for by the following heuristic arguments based on the peculiar architecture of the networks . at first sight , this family of modular networks is not a very compact system , since in these networks , nodes with large degrees are not directly linked to one another , but connected to those nodes with small degree . however , this network family is made up of many small densely interconnected clusters , which combine to form larger but less compact groups connected by nodes with relatively high degrees . for node pairs in a small group , their shortest path length is very small because of the high cohesiveness of small modules . for the length of shortest paths between two nodes belonging to different large groups , it seems long because the groups that the nodes lie at are not adjacent to each other . but this is not the fact . by construction , although the relatively large groups are not directly adjacent , they are joined by some large nodes , which are connected to each other by a layer of intermediate small - degree nodes ( see figure [ network ] ) , such as the peripheral nodes or locally peripheral nodes @xcite . thus , different from conventional random scale - free networks , especially assortative networks @xcite , in the studied scale - free modular networks , although large - degree nodes are not connected to one another , they play the role of bridges linking different modules together , which is the main reason why the average distance of the networks is small . it deserves to be mentioned that , although the studied modular scale - free networks display small - world behavior , the logarithmic scaling of average distance with respect to network order is different from the sublogarithmic scaling for conventional non - modular stochastic scale - free networks with degree distribution exponent @xmath92 , in which the average distance @xmath93 behaves as a double logarithmic scaling with network order @xmath94 , namely , @xmath95 @xcite . thus , despite that the degree distribution exponent of the modular scale - free networks is smaller than 3 , their average distance is larger than that of their random counterparts with the same network order . the root of this difference may also lie with the modular structure , particularly the indirect connection of large nodes , as addressed above . the genuine reasons for this dissimilarity need further studies in the future . the determination and analysis of average distance is important to understand the complexity of and dynamic processes on complex networks , which has been a subject of considerable interest within the physics community . in this paper , we investigated analytically the average distance in a class of deterministically growing networks with scale - free behavior and modular structure , which exist simultaneously in a plethora of real - life networks , such as social and biological networks . based on the self - similar structure of the networks , we derived the closed - form expression for the average distance . the obtained exact solution shows that for very large networks , they are small - world with their average distance increasing as a logarithmic function of network order . we confirmed the rigorous solution by using extensive numerical simulations . we also showed that the small - world behavior lies with the inherent modularity and scale - free property of the networks . we would like to thank xing li for his support . this research was supported by the national natural science foundation of china under grants no . 60704044 , no . 60873040 , and no . 60873070 , the national basic research program of china under grant no . 2007cb310806 , shanghai leading academic discipline project no . b114 , the program for new century excellent talents in university of china ( grants no . ncet-06 - 0376 ) , and shanghai committee of science and technology ( grants no . 08dz2271800 and no . 09dz2272800 ) .	various real - life networks of current interest are simultaneously scale - free and modular . here we study analytically the average distance in a class of deterministically growing scale - free modular networks . by virtue of the recursive relations derived from the self - similar structure of the networks , we compute rigorously this important quantity , obtaining an explicit closed - form solution , which recovers the previous result and is corroborated by extensive numerical calculations . the obtained exact expression shows that the average distance scales logarithmically with the number of nodes in the networks , indicating an existence of small - world behavior . we present that this small - world phenomenon comes from the peculiar architecture of the network family .
transient radio sky or time - variable radio sources in space have been recognized as one of the key science drivers for the square kilometre array ( ska ) @xcite what will be reflected in its design and operations @xcite . this science area is marked as `` exploration of the unknown '' what includes the likely discovery of new classes of objects and phenomena . yet the statistical properties of transient radio sky remain largely unknown , what includes even the high - energy transients as they seem to be unfolding on very short time scales . in detection approach for the arrays such as mwa , lofar and ska we propose to acknowledge the need for reducing data transport and computational cost while searching for fast transients . therefore , we suggest that such detection needs to be done in stages with reduced data for storing and transporting at every consequent stage . firstly , the presence of the abrupt change in raw time - domain data needs to be identified . secondly , the evidences of the extraterrestrial nature of the detected signal need to be found . thirdly , if such evidence is present , more detailed analysis can be applied . in this paper we present an attempt to develop a reliable and cost - effective technique for the first stage of detection using statistical signal detection approach . the proposed method of handling received time domain sequences is based on online abrupt changes detection scheme . the algorithm operates in time domain and is required to identify the onset of the signal by checking if the threshold of the detection process has been reached . let s consider the signal with the following structure : @xmath0 where _ t _ is time , _ ( _ t _ ) @xmath1 is a gaussian noise process with mean @xmath2 and variance @xmath3 , _ s_(_t _ ) @xmath4 is a gaussian signal with mean @xmath5 and variance @xmath6 . let @xmath7 and @xmath8 since we assume that signal and noise are independent . @xmath9 and @xmath10 are hypothesis under which the signal of interest is either absent ( hypothesis @xmath9 ) or present ( @xmath10 ) along with the additive background noise n(t ) . as we demonstrate further , the proposed utilization of algorithm known as cumulative sum ( cusum ) introduced by @xcite allows for mean only parameter to be sufficient when building a detector . changes detection scheme , used ino transient identification algorithm , is based on two generic detection procedures : log likelihood ratio ( llr ) and cusum . llr test is based on using a ratio of two @xmath11 to build an indicator upon which a threshold can be applied . if the ratio exceeds given threshold , it indicates the prevalence of one of the hypothesis over another . lr for a gauss - distributed data using the model given in ( [ eq2 ] ) is : @xmath12 we consider identification successful when @xmath13 , where @xmath14 is a predefined threshold for identification . llr is obtained by taking logarithm of ( [ eq4 ] ) : @xmath15 equation ( [ eq5 ] ) can now be expressed through either mean- or variance - based detector . in the latter case , assuming equal and zero mean , we obtain : @xmath16 assuming equal and unity variance , ( [ eq5 ] ) becomes : @xmath17 . thus , defining @xmath18 and rescaling the log likelihood expression by the proportionality value ( @xmath19 ) to obtain a threshold value . if the probability of type i error @xmath20 is fixed , the detection threshold can be expressed as @xmath21 . cusum is a detection procedure proposed by @xcite . cusum is a repeated llr test for a change from one known distribution to another . we assume that for known densities @xmath22 and @xmath23 there exists an unknown change point @xmath24 , where the input sequence @xmath25 if @xmath26 and @xmath27 if @xmath28 . the cusum statistics is the one that satisfies the following recursion : @xmath29 where @xmath30 is a likelihood ratio for @xmath31 evaluated at @xmath32 . the procedure raises an alarm at time @xmath33 for threshold @xmath34 , which is discussed later . if likelihood value from ( [ eq5 ] ) used with regard to variances , the cusum recursion would take the following form , based on the usual radiometric output being the averaged sum of the samples squares or the energy detector : @xmath35 with the parameter @xmath36 denoting variance - based coefficient @xmath37 . since it relies on the full knowledge of variances of the signal and noise , it is not always practical to use cusum algorithm with the recursion denoted in ( [ eq21 ] ) . applying ( [ eq13 ] ) cusum may be expressed as a procedure , which starts with @xmath38 , it recursively calculates : @xmath39 and stops as soon as @xmath40 exceeds threshold @xmath34 . generally speaking , equation ( [ eq23 ] ) can be rewritten as @xmath41 , where @xmath36 being a configurable parameter . choice of this parameter is ruled by the expected behavior of the procedure , as stated by @xcite : `` scoring is chosen so that the mean sample path on the chart when quality is satisfactory is downwards and is upwards when quality is unsatisfactory '' . as demonstrated by @xcite the optimal value for parameter r will be the largest acceptable mean value or the smallest unacceptable mean value . one of the benefits of cusum scheme for signal detection is its stability in the presence of regression behavior in input data @xcite , which is due to the gradual changes in the system temperature or changes of the overall sky temperature for large field of view or `` all - sky '' telescopes . choice between mean - based and variance - based cusum schemes depends on the type of the receiver used . for a linear output receiver , sampled according to nyquist , the output represents voltage changes with time . both mean- and variance - based indicator function can be used . however , mean - based indicator is preferred because it does not rely on variance - based separation of noise and signal . for full power square - law detector , however , mean - based indicator function becomes unusable due to the loss of information on mean . therefore , variance - based cusum scheme should be used discounting the utilization of mean @xcite . cusum scheme requires a threshold value @xmath34 to be chosen , which , when crossed , identifies the point of abrupt change in the statistical characteristics of the signal . threshold value @xmath34 can be derived from wald test on the mean of a normal population @xcite . assuming that @xmath42 , the value of threshold @xmath34 is equivalent to a sequence of wald sequential tests @xcite with boundaries @xmath43 : @xmath44 where @xmath20 was interpreted by @xcite as a crude approximation to the proportion of samples that trigger false alarm . this value can also be interpreted as a probability of false alarm in a traditional sense . since we assumed that @xmath45 ( [ eq_threshold ] ) can be rewritten as @xmath46 . the algorithm was tested on the data obtained from parks radiotelescope . input data contained 1 second observation from the vela pulsar psr j0835 - 4510 . the observation was sampled at 1416 mhz with 64 mhz bandwidth . psr j0835 - 4510 has a period of 89 ms . figure [ fig : vela_det ] represents a portion of the observed data with the detected pulses marked by vertical red lines , with blue lines representing 0.1 second intervals . out of 10 impulses contained in the test data , 9 were identified . 90% of pulses being correctly identifed prove the applicability of cusum procedure described above upon the raw , time - domain radio data . detectability of a signal is controled by and relied upon the preset threshold , which is calculated based on the assumed probability distribution of data ( [ eq_threshold ] ) . , observed at parks telescope , central frequency @xmath47 mhz , bandwidth @xmath48 mhz , recorded with 2-bit vlbi recorder . the recording started at @xmath49 . short bars mark @xmath50 intervals . vertical red lines mark the beginning of detected pulses . , width=623,height=136 ] we have presented the algorithm , which provides reliable and computationally efficient detection of dispersed radio transients in time domain . the algorithm is well - suited for being implemented on fpga or gpu platforms for real time detection on arrays such as ska . statistical methods used in the algorithm provide easy staging of detection process across multiple handling points giving the opportunity for significant reduction of data volume at each consequent stage of detection . the authors would like to thank aidan hotan of curtin institute of radioastronomy for making @xmath51 pulsar data available to us . basseville , m. , & nikiforov , i. , _ detection of abrupt changes : theory and application _ , englewood cliffs , nj , prentice hall , 1993 box , g. , & ramirez , j. , _ cumulative score charts _ , quality and reliability international , 8,17 - 27 , 1992 chang , j. t. , & fricker , r.d . jr , _ detecting when a monotonically increasing mean has crossed a threshold _ , journal of quality technology , 31,2 , 217 , 1999 cordes , j. m. , _ the square kilometre array as a radio synoptic survey telescope : widefield surveys or transients , pulsars and eti _ , ska memo . ska , 2007 ( rev . fridman , p. , _ a method of detecting radio transients _ , http://arxiv.org/pdf/1008.3152 , 2010 gan , f. f. , _ cusum control charts under linear drift _ , the statistician , 41 , 71 , 1992 koenig , r.,_candidate sites for world s largest telescope face first big hurdle _ , science , 313 , 5789 , 910 , 2006 lorden , g. , _ procedures for reacting to a change in distribution _ , annals of mathematical statistics , 42 , 1897 , 1971 manly , b. f. j. , _ the choice of a wald test on the mean of a normal population _ , biometrika , 57 , 1 , 91 , 1970 moustakides , g. , _ optimal stopping times for detecting changes in distributions _ , annals of mathematical statistics , 14 , 1379 , 1986 page , e. , _ continuous inspection schemes _ , biometrika , 41 , 100 , 1954 van dobben de bruyn , c.s . , _ cumulative sum tests - theory and practice _ , hafner publishing co , new york , 1968	computationally inexpensive algorithm for detecting of dispersed transients has been developed using cumulative sums ( cusum ) scheme for detecting abrupt changes in statistical characteristics of the signal . the efficiency of the algorithm is demonstrated on pulsar psr j0835 - 4510 . [ firstpage ]
within the local supercluster , multi - color ccd surface photometric data are now available on a number of flux - limited samples of disk galaxies in clusters ( e.g. , pierce & tully 1988 ; tully et al . 1996 ) , but the same is not true on field disk galaxies . lu et al . ( 1993 ; hereafter paper i ) selected , in two separate volumes in the local supercluster ( hereafter lsc ) , all the disk galaxies brighter than @xmath2mag in the uppsala general catalogue of galaxies ( hereafter ugc ; nilson 1973 ) . one of the volumes is toward but beyond our local group as viewed from the virgo cluster . with a median heliocentric velocity of about 2000 , the ugc sample in this `` anti - virgo cluster '' volume is located in the outer part of the local supercluster where environmental effects on galaxy disks are probably much less important than in and near the virgo cluster , but still close enough to us to have a fairly faint absolute magnitude limit of @xmath3mag . as a result , this sample is particularly suitable for probing the statistical properties of the disk galaxy population . ccd surface photometry in the i - band and high s / n - ratio 21 cm hi data on this sample are already published in paper i. in this paper , we further present new ccd surface photometry in the b - band on this sample . as an application of this data set , the b - band data are used in the second half of this paper to probe the distribution function of galaxy disk parameters . an exponential disk can be fully described by two fundamental parameters : a central surface brightness ( hereafter csb ) and an ( exponential ) disk scale length ( hereafter dsl ) . how disk galaxies are distributed in terms of these two parameters is important as it may carry information about the physical condition of the universe at the galaxy formation epoch ( e.g. , dalcanton et al . so far , efforts have been mostly in determining this distribution function partially integrated over dsl , namely , the csb distribution . freeman ( 1970 ) showed that the csbs of disks of a sample of local disk galaxies are distributed in a narrow range of @xmath4 . this so - called freeman law was later interpreted by disney ( 1976 ) and disney & phillipps ( 1983 ) to be a selection effect due to the fact that a sample selected on the basis of a limiting diameter at a fixed surface brightness ( hereafter sb ) may miss giant disk galaxies ( with a large dsl ) whose csbs are too faint , as well as compact galaxies ( with a bright csb ) whose dsls are too small . note that the same selection effect could also occur in a flux limited sample . the existence of the above selection effect in optical galaxy catalogs such as ugc has been supported by a large number of studies ( e.g. , allen & shu 1979 ; davies 1990 ; schombert et al . 1992 ; mcgaugh et al . 1995 ; mcgaugh 1996 ; de jong 1996 ; sprayberry et al . 1996 ; dalcanton et al . 1997b ; and the references in bothun et al . but there is still considerable controversy about the exact shape of the csb distribution for relatively bright disk galaxies . for example , using a diameter - limited galaxy sample and a volume sampling function based on both csb and dsl , van der kruit ( 1987 ) concluded that after excluding dwarf galaxies , there are not many large , low - csb disk galaxies ; on the other hand , using larger galaxy samples and a volume sampling function based on csb alone , other groups have derived csb distributions that are nearly flat at values fainter than the canonical freeman s value of [email protected] ( e.g. , davies 1990 ; mcgaugh et al . 1995 ; mcgaugh 1996 ) . for a further discussion on this controversial subject , see briggs ( 1997 ) . there are , however , a number of issues which have not been formally addressed in the past and which may be important to correctly interpreting any csb distribution function derived from a diameter or flux limited sample . the first one concerns the so - called malmquist bias ( malmquist 1920 ) . at a given csb , the dsl distribution is fairly wide ( e.g. , mcgaugh et al . 1995 , de jong 1996 ) . a diameter ( or flux ) limited selection leads to a malmquist - type bias in the sense that one tends to select only intrinsically large disks at a given csb . near the limiting sb of a sample selection , a slight dimming in csb has to be compensated by a large increase in dsl in order for a galaxy to be selected . therefore , a bias could occur between high- and low - csb disk galaxies . the second issue is how to take into account the effect of disk inclination on the detectability of a disk galaxy . depending on how transparent a galaxy disk is in optical , this effect could be particularly important to galaxies with disk parameters near the selection threshold . as we show in this paper , depending on whether disks are opaque or transparent , the same galaxy sample could lead to a quite different conclusion on the derived distribution function of galaxy disk parameters . the third issue regards whether the variables , csb and dsl , are separable in the bi - variate galaxy distribution function . in other words , are csb and dsl statistically independent of each other ? a positive answer to this question would allow one to derive a csb distribution function from a diameter or flux limited sample without requiring complete redshift data . some authors have argued for and used such a statistical independence between csb and dsl ( e.g. , mcgaugh 1996 ) , but this has never been rigorously tested . a possible correlation between the two parameters is hinted by theoretical considerations ( e.g. , dalcanton 1997a ) and by the observational fact that low - csb spiral galaxies tend to have a large dsl ( e.g. , kent 1985 ; bothun et al . 1990 ; dalcanton et al . when examining the apparent dsl distribution of galaxies from a diameter or flux limited sample , one has to be aware of the malmquist bias : at fixed csb and distance , the bias prevents one from sampling galaxies with a dsl shorter than some threshold . without a prior knowledge of the intrinsic dsl distribution function , the only secure measurement one can make is on the part of the dsl distribution above this threshold . therefore , for a set of well defined selection criteria , and as we show in this paper , one can still answer the question as to whether the dsl distribution at large values depends on csb . the remainder of this paper is organized as follows : in sect . 2 , we describe the galaxy sample , our b - band ccd surface photometry and present the photometric results . in sect . 3 , we illustrate the two selection effects introduced above ; and by taking into consideration these selection effects , present a method of constructing a volume sampling function and a way to interpret the derived distribution function of disk parameters . the method is then applied to our galaxy sample . in sect . 4 , we discuss some implications from our analyses . we end this paper with a brief summary in sect . 5 . throughout this paper , we assume a hubble constant of 75 , and use the following notations for an exponential disk : @xmath5 ( @xmath6 ) for its observed ( face - on ) csb in units of and @xmath7 ( @xmath8 ) for its angular ( linear ) dsl in units of arcsec ( kpc ) . thus , an exponential disk can be expressed as @xmath9 our galaxy sample , originally selected in paper i and used in lu et al . ( 1994 ) to study the local velocity field , contains all the 76 ugc disk galaxies with @xmath10 mag within a volume bounded by @xmath11 , @xmath12 and a heliocentric velocity of @xmath13 . our b - band ccd observations were carried out with the hale @xmath14 telescope equipped with the four shooter ( gunn et al . 1987 ) at palomar observatory from august 25 to 27 , 1990 ( ut ) under a photometric condition . we used the 4-shooter s standard violet filter ( 4300/700 ) to mimic johnson s b system ( johnson & morgan 1953 ) . the resulting ccd field is a @xmath15 square with a pixel size of @xmath16 . in addition to this ugc sample , we also observed a number of optically fainter galaxies as described in paper i ( also see hoffman et al . the integration time per galaxy ranges from 4 to 8 minutes . the data reduction procedure is similar to that in paper i. the final images ready for surface photometric analysis show a quite flat background . for example , the mode and mean of sky pixels usually agree with each other within 0.3% . the instrumental magnitudes were converted into johnson b system using the observed standard stars taken from landolt ( 1983 ) . surface photometric analysis was done by fitting elliptical contours to each sky - subtracted galaxy image following the prescription given in paper i. the fitting was performed on the b - band images , independent of the existing i - band results in paper i. in most cases , a good fit could be obtained down to a sb of b@xmath17 . for each galaxy , the resulting sb profile as a function of the semi - major axis , @xmath18 , was displayed and its outer part between two radii @xmath19 and @xmath20 , dominated by the disk component as judged by eyes , was fit into eq . . the mean ellipticity of the disk component , @xmath21 , was evaluated between the radii @xmath19 and @xmath20 . the isophotal magnitude , @xmath22 , and diameter , @xmath23 , were measured at b@xmath126 isophote determined by the fitted exponential disk profile . for a number of galaxies whose b@xmath17 isophotes are partially outside the ccd field , the exponential disk fit was used in each case to evaluate the contribution to @xmath22 from those isophotes partially outside the ccd field . finally , the total magnitude , @xmath24 , was evaluated by extrapolating the isophotal magnitude at @xmath20 to infinity in radius using the exponential disk fit . we list in table 1 all the 76 ugc sample galaxies as well as those optically fainter ones that we observed . of the 76 ugc sample galaxies , seven do not have photometric data for various reasons as given at the end of table 1 . the table columns are as follows : col . ( 1 ) is the galaxy name as in ugc , but for those fainter galaxies we give their names as in paper i. col . ( 2 ) gives the ngc or ic number if applicable . the adopted distance in mpc is given in col . ( 3 ) , derived from the velocity of the galaxy with respect to the centroid of the lg [ i.e. , @xmath25 . cols . ( 4 ) and ( 5 ) are @xmath19 and @xmath20 in arcsec , respectively ; namely , the inner and outer radii for the exponential disk fit . col . ( 6 ) is the mean disk ellipticity , @xmath21 , which has been used to derive the disk inclination angle in degrees in col . ( 7 ) via @xmath26 col . ( 8) is the mean position angle of the disk on the sky ( n to e ) measured between the radii @xmath19 and @xmath20 ( 9 ) and ( 10 ) are respectively @xmath22 in mag and @xmath23 in arcmin . ( 11 ) is the angular dsl in arcsec and col . ( 12 ) the linear dsl in kpc . ( 13 ) is @xmath5 in determined from the exponential disk fit . ( 14 ) is the total magnitude @xmath24 . ( 15 ) is the b - band absolute magnitude derived from @xmath27 and the distance in col . ( 3 ) . finally , col . ( 16 ) gives @xmath28 color derived from @xmath24 in this paper and @xmath29 in paper i. no correction for galactic or internal reddening has been applied to the parameters in table 1 . 1 displays the observed b - band surface brightness as a function of the semi - major axis for each of the galaxies with photometric parameters in table 1 . the open squares represent the measured isophotes , while the filled square represents the fitted isophote at b@xmath17 . a number of galaxies were observed multiple times over the entire observing run . the multiple images of the same galaxy were reduced independently from each other and the results are used as a way to measure the statistical uncertainties in the derived photometric parameters . such estimated r.m.s . uncertainties are on the order of @xmath30 mag for @xmath22 , @xmath31 mag for @xmath24 , @xmath32 for the disk inclination angle , @xmath33 for @xmath5 , and @xmath34 for @xmath7 . another way to illustrate our photometric accuracy is to compare the b - band result here with the i - band result in paper i on the same galaxy . as an example , plotted in fig . 2 as functions of the b - band disk ellipticity are the differences in the measured disk position angle ( p.a . ) and ellipticity between the two bandpasses . as expected , the more inclined a galaxy disk is , the better agreement between the two bandpasses is in fig . 2 . for galaxies inclined more than @xmath35 ( @xmath36 ) , the typical agreement between the two bandpasses is within @xmath37 in p.a . and @xmath38 in disk ellipticity ( or @xmath39 in terms of disk inclination angle ) . our @xmath24 magnitudes are however fainter by about 0.06 mag on average than the @xmath40 magnitude scale of the third reference catalogue of galaxies ( rc3 ; de vaucouleurs et al . this is illustrated in fig . 3 . a gaussian curve , with a center at @xmath41 mag and a fwhm of @xmath42 mag , is shown in the figure for comparison . no obvious correlation could be identified between @xmath43 and the night on which @xmath24 was obtained , galaxy morphology , optical color or @xmath40 . it is also interesting to see how the fraction of light outside the b@xmath126 isophote vary with the disk central surface brightness . 4 shows that @xmath44 increases as @xmath5 increases . note that , for a low - sb galaxy of @xmath45 , more than half of its luminosity lies outside the b@xmath17 isophote . because we did not do a full bulge / disk decomposition , the csb of a galaxy with a prominent bulge could be overestimated ( kormendy 1977 ) . we found however that the eraly - type disk galaxies do not show on average a brighter csb than those late - type galaxies , suggesting that our disk fitting procedure is probably insignificantly affected by the size of a galactic bulge . on the other hand , there are a number of sample galaxies with prominent spiral arms forming a ring - like pattern . these include ugc@xmath112343 , ugc@xmath112447 , ugc@xmath112777 , and ugc@xmath100858 . for each of these galaxies , the sb profile outside the spiral arms , where we have fit its exponential disk , radially falls off fairly quickly to a faint level . the resulting csbs of these galaxies are among the brightest in the sample . should we have fit an exponential disk to the entire galaxy surface , we would have obtained a fainter csb in each galaxy . it is not clear which way is better . but not all high - sb sample galaxies are of this type . for example , ugc@xmath112074 , ugc@xmath112529 and ugc@xmath100167 are also among those of the brightest csbs in the sample , but none of them show prominent spiral arms in optical . in fact , with @xmath46 and a moderate disk inclination , these 3 disk galaxies may represent a class of relatively rare , `` super high - sb '' disk galaxies . we will study these three galaxies in more details in a future paper . as a conventional simplification , we formulate the ugc sample selection by assuming a negligible effect from the bulge of a galaxy . this is probably a reasonable simplification as most of our sample galaxies are dominated by their disks . we define an intrinsic or face - on csb as follows : @xmath47 the transition at @xmath48 in eq . ( 3 ) is chosen somewhat arbitrarily . it corresponds to the onset of @xmath49 when the disk inclination angle @xmath50 is given by eq . the value of @xmath51 is limited to @xmath52 , with the lower limit corresponding to a completely opaque disk and the upper limit to a fully transparent disk . in spite of extensive studies on the opacities of galaxy disks , it is still highly controversial as to whether galaxy disks are largely opaque or transparent ( e.g. , tully & fouqu 1985 ; disney , davies , & phillipps 1989 ; valentijn 1990 , 1994 ; burstein , haynes , & faber 1991 ; byun 1993 ; giovanelli et al . 1994 ; xu & buat 1995 ; tully & verheijen 1997 ) . in this paper we consider only two cases : ( a ) fully transparent disks with @xmath53 and ( b ) fairly opaque disks with @xmath54 . we plot in fig . 5 @xmath5 as a function of @xmath7 for the 69 ugc sample galaxies with b - band photometry . the filled and open squares represent those with a disk ellipticity below and above @xmath55 , respectively . it is clear that , at a given @xmath7 , disks of larger inclination angles are on average associated with brighter values of @xmath5 [ cf . inequality ( 6 ) below ] . the solid curve in the figure represents the selection limit associated with the ugc limiting diameter of @xmath56 at @xmath57 ( cornell et al . 1987 ) as follows : @xmath58 for an exponential disk , its total magnitude can be written as @xmath59 our magnitude selection criterion of @xmath60 mag transfers to @xmath61 note that as in eq . ( 3 ) , we simply set @xmath48 in both criteria ( 5 ) and ( 6 ) for cases of @xmath62 . it is clear that only for a fully transparent disk , is criterion ( 5 ) independent of @xmath21 . we plot criterion ( 6 ) in fig . 5 for the cases of @xmath63 and @xmath48 by the dotted and dashed curves , respectively . to have a rough , but quantitative picture of the overall sample selection , we give in table 2 a few numerical indicators on how our sample selection acts on _ face - on _ disks of a given csb : column ( 2 ) is the minimum angular dsl for a galaxy of a given csb to be selected . this is given by combining criteria ( 4 ) and ( 6 ) . column ( 3 ) gives @xmath64 , where @xmath65 is the sample distance cutoff and is taken to be 42.5 mpc in this paper , and @xmath66 is the maximum distance at which a galaxy of @xmath67mag can still be selected . this quantity is given by relating the minimum @xmath7 in column ( 2 ) to the following relation on the absolute magnitude in the case of @xmath67mag and @xmath63 , @xmath68 column ( 3 ) shows that a disk galaxy of @xmath69 mag can be seen up to about half of @xmath65 in distance for the most part of the csb range explored here . so roughly speaking , the part of the galaxy population with @xmath70mag is adequately sampled here . finally , columns ( 4 ) and ( 5 ) are @xmath71 and @xmath72 , respectively , where @xmath71 is the linear dsl of a galaxy of @xmath67mag via eq . ( 7 ) and @xmath72 is the minimum @xmath8 that a disk galaxy has to have in order to be selected out to the sample cutoff distance @xmath65 . the presence of a malmquist bias in our sample is illustrated in fig . 6 , where we plot @xmath73 as function of @xmath8 for the sample galaxies in two cases : ( a ) fully transparent disks with @xmath74 , and ( b ) fairly opaque disks with @xmath75 . the dotted and solid curves in the figure are generated by using columns ( 4 ) and ( 5 ) of table 2 , respectively . clearly , the distribution pattern of the data points in each plot suggests that at a given csb , only galaxies with large enough @xmath8 have been selected . this bias becomes progressively severe when the csb under consideration approaches the limiting sb of the sample selection . without taking into consideration this bias , a distribution of csb or dsl derived from our ugc sample would also be biased . unfortunately , one usually does not have a prior knowledge about the intrinsic shape of the @xmath8 distribution , especially at a faint csb level , a correction for this malmquist bias remains model dependent at best . although this malmquist bias makes it impossible to use our sample to conduct a full bi - variate analysis of the galaxy distribution , we can still determine , at a given csb , the part of the @xmath8 distribution that is adequately sampled by our sample . roughly speaking , this is the region to the right of each dotted curve in fig . let us consider a galaxy of an exponential disk with fixed @xmath73 and @xmath8 . the corresponding observables are @xmath5 and @xmath7 , respectively . such a galaxy would be selected if its distance @xmath76 , which relates @xmath8 to @xmath7 , and inclination angle @xmath50 , which relates @xmath73 to @xmath5 via eq . ( 3 ) , are such that both criteria ( 4 ) and ( 6 ) are satisfied . we sketch this in fig . 7 for both the cases of @xmath77 and @xmath75 , with the help from the same curves as shown in fig . 5 . for a fully transparent disk with @xmath74 , at a given @xmath7 , an increasing disk inclination would move the galaxy vertically upward , as illustrated by the thick arrow , from the horizontal line marked as `` @xmath78 '' to the one marked as `` @xmath79 ( k=1.0 ) . '' this remains true as long as @xmath7 is greater than that of the vertical line `` a - b@xmath80 '' in the figure , which marks the farthest point in distance at which this galaxy can still be selected in our ugc sample . for a fairly opaque disk with @xmath75 , at a large value of @xmath7 , an increasing disk inclination still moves the galaxy vertically upward , but only to the horizontal line marked as `` @xmath81 ( k=0.2 ) . '' the situation changes when the @xmath7 value of the galaxy becomes smaller than that of point `` b2 '' in fig . when this happens , at each @xmath7 , the galaxy can move vertically up to the line `` a - b@xmath82 '' as illustrated by the thin arrow in fig . 7 . in other words , the galaxy would be selected only if its disk inclination is small enough , and at point `` a '' the galaxy would be included in our ugc sample only if its disk is viewed face - on . to summarize , the detectability of an optically thin disk is much independent of its inclination , while a fairly opaque disk could be selected farther in distance at smaller disk inclination angle . this statement needs a slight modification when the solid curve surpasses both the dashed and dotted curves in fig . 7 and becomes the most stringent selection criterion , i.e. , at @xmath83 or @xmath84 . but as evident in fig . 5 , few sample galaxies are in these regimes . we now incorporate this inclination dependent detectability into a volume sampling function . for a given galaxy at a given distance @xmath76 ( or @xmath7 ) , one can define @xmath85 , the maximum `` volume '' in the phase space of the disk inclination angle , to be @xmath86 where @xmath87 and @xmath88 , both in degrees , are the minimum and maximum possible values for the inclination angle of this galaxy as described above . @xmath87 can be either zero or greater . for @xmath53 , @xmath88 always equals @xmath89 . for @xmath90 , @xmath88 could be either @xmath91 or smaller . next we let the distance of the galaxy vary ( so does @xmath7 ) , both @xmath87 and @xmath88 are now functions of the distance of the galaxy . denote @xmath92 as the maximum distance a galaxy can still be detected when it is at the most favorable disk inclination ( e.g. , point `` a '' in fig . 7 ) , we can define a composite maximum ( space @xmath93 inclination ) phase volume as follows @xmath94 where @xmath95 is the constant solid angle ( @xmath96sr ) of our sample on the sky , and @xmath65 ( @xmath97mpc ) is our sample distance cutoff . let @xmath98 be the space density distribution function of disk galaxies in terms of csb and dsl , in the absence of the malmquist bias , one would have @xmath99 where the sum is over all the sample galaxies with @xmath73 and @xmath8 within the intervals @xmath100 and @xmath101 , respectively . we have calculated @xmath102 for each of the 69 ugc sample galaxies with the b photometry in table 1 . although we left out the other 7 sample galaxies because of their unavailable b photometry , this should not have a significant effect on the shape of our derived distribution function . the malmquist bias and our moderate sample size prevent us from using eq . ( 10 ) directly . instead , we divide the ugc sample into 4 consecutive bins in the inclination corrected csb and derive a dsl distribution within each of these bins . the results are shown in fig . 8 for the case of @xmath53 and in fig . 9 for the case of @xmath54 . one galaxy ( ugc@xmath101466 ) with @xmath103kpc and a relative density of @xmath104 for the bin of @xmath105 is off the figure . as noted in table 1 , the disk fit of this galaxy was performed over a range of radii dominated by the prominent spiral arms of the galaxy . it is likely that its dsl has been overestimated . a total of 10 sample galaxies with @xmath106mag have also been excluded from these figures . should they be included here , most of them would occupy the region in each csb bin to the left of the arrow which roughly indicates the threshold in @xmath8 below which the malmquist bias makes the distribution incomplete ( see fig . 6 or table 2 ) . the part of the distribution to the right of the arrow is considered here to represent the true dsl distribution subject to the statistical error . it is this part of the distribution we use to draw the following results . let us concentrate on the two faintest , equally wide csb bins in these figures , namely , ( i ) @xmath105and ( ii ) @xmath107 . these bins cover the flat part of the csb distribution shown , for example , in mcgaugh ( 1996 ) . in the case of transparent disks with @xmath74 ( see fig . 8) , the distribution in the fainter csb bin ( ii ) has a longer tail toward large @xmath8 values . for example , the integrated density over @xmath108kpc is about @xmath109(@xmath110 ) in bin ( ii ) . note that all the galaxies with @xmath108kpc in bin ( i ) would be detectable up to the maximum sample distance ( cf . table 2 ) . should the csb distribution be nearly flat with csb and dsl being statistically independent of each other ( e.g. , mcgaugh 1996 ) , one would expect to see about @xmath111(@xmath112 ) [ @xmath113 galaxies with @xmath108kpc in bin ( i ) . but we actually observed only one galaxy . thus , at a significance level of 2.5@xmath114 , our analysis suggests either ( a ) that csb and dsl are correlated in the sense that the dsl distribution at a fainter csb level has a longer tail toward large values ; or ( b ) that csb and dsl are still independent of each other , but with a csb distribution function that increases rapidly toward faint csb values ( i.e. , at a rate 10 times faster than a flat csb distribution ) . in the case of fairly opaque disks with @xmath54 as shown in fig . 9 , we have only marginally useful data in the faintest csb bin ( ii ) . under the assumptions of a nearly flat csb distribution and a statistical independence between csb and dsl , the same analysis as above implies that the expected number of galaxies with @xmath108kpc in the csb bin ( i ) is @xmath115(@xmath116 ) . we actually observed 5 galaxies in that bin . however , this comparison is only significant at 1.5@xmath114 . the two selection effects that we discussed and formulated in the previous section , namely , a malmquist bias and a disk inclination dependent detectability , should be present in any diameter or / and flux limited sample . as we have shown here , both of these could have a significant effect on how to interpret a distribution function derived from a diameter / magnitude limited sample . we note that neither the visibility theory of disney and phillipps ( 1983 ) nor the volume sampling function of mcgaugh ( 1996 ) has fully addressed these selection effects . for a diameter limited sample , the volume sampling function is proportional to @xmath117 ( mcgaugh 1996 ) , where @xmath118 is the limiting sb in the sample selection . this sensitive dependence on @xmath8 makes the malmquist bias particularly severe near the limiting sb of a sample selection . for example , our fig . 6 shows that @xmath119 , the threshold in @xmath8 below which our ugc sample is highly incomplete , increases from about 1.5@xmath1kpc for the range of @xmath120 to 3.5@xmath1kpc for @xmath121 . this malmquist bias , if not corrected for , has the following implications : if the full volume sampling function as defined above is used ( see an example in van der kruit 1987 ) , one will always overestimate the _ mean _ volume sampling function for the population of the low - csb galaxies relative to that of the high - csb galaxies , leading to a relative underestimate of the galaxy number density at low csb values . on the other hand , if a volume sampling function that scales only with @xmath122 is used ( see an example in mcgaugh 1996 ) , one will relatively underestimate the volume sampling functions for the low - csb galaxies in the sample . although this underestimate works in favor of compensating the malmquist bias , it is no guarantee that the compensation would work out perfectly . to better understand how disk parameters are distributed at the low sb end and to effectively constrain disk formation models such as that proposed by dalcanton et al . ( 1997a ) , we need galaxy samples selected with a small limiting diameter at a faint sb . unfortunately , this implies that the resulting catalog might be too large for achieving a completeness in redshift . one alternative way is to use some cluster sample as illustrated in tully & verheijen ( 1997 ) , for which complete redshift data are not needed ( except for weeding out field galaxy contaminations ) . note that the selection effects discussed in this paper still apply to a cluster sample . a good example of this is shown in fig . 4 of dalcanton et al . ( 1997a ) on a complete sample of virgo cluster galaxies . a systematic hi survey ( e.g. , szomoru et al . 1994 ) with follow - up optical ccd photometry on the detected galaxies is another alternative approach that could offer an unbiased picture on how galaxy disk parameters distribute above a certain threshold in hi mass . although the shape of the csb distribution function is still quite uncertain at faint sb levels , it is rather clear from this study and those cited in sect . 1 that this distribution function is fairly wide over csb even for bright disk galaxies , extending to much fainter values than the narrow range initially proposed by freeman . however , our data ( see figs . 5 and 6 ) do not show a clear support for a bimodal csb distribution . this is further confirmed by our i - band data from paper i in which disk internal reddening is only about 40% of that in the b - band . a bimodal csb distribution is observed on disk galaxies in the ursa major cluster in both i and k ( tully & verheijen 1997 ) . it is not clear at this point if this difference implies that the bimodal csb distribution is a phenomenon specific to certain clusters . in this paper , we present new b - band ccd surface photometry on ( 1 ) 69 galaxies in a complete sample of 76 disk galaxies brighter than @xmath0mag in the uppsala general catalogue of galaxies and ( 2 ) 11 additional galaxies that are optically fainter than 14.5@xmath1mag . surface brightness and color radial profiles are shown and various photometric parameters are tabulated on each of these galaxies . this data set complements our earlier published i - band ccd and high s / n - ratio 21 cm hi data on the same galaxies ( lu et al . 1993 ) . the b - band data are then used to study the distribution of the fundamental galaxy disk parameters : the central surface brightness ( csb ) and ( exponential ) disk scale lengths ( dsl ) . we illustrate two selection effects that occur in any diameter or / and flux selected sample of disk galaxies : ( 1 ) there is always a malmquist - type selection effect that biases against disk galaxies with small disk scale lengths at a given csb , and ( 2 ) there could be a dependence of the detectability of a galaxy on it disk inclination angle as long as disks are not completely opaque . without a prior knowledge on the full distribution function of the disk parameters , it is difficult to fully correct for the malmquist bias . on the other hand , we derive a volume sampling function that takes into account the inclination effect . using this volume sampling function , we derive a relative density distribution of dsl for a given range of csb values from the ugc sample for each of the following two cases : ( a ) fully transparent disks and ( b ) fairly opaque disks . replying on only the part of the resulting distribution function that is least affected by the malmquist bias , we show that in the case of ( a ) , csb and dsl could be correlated in the sense that , up to an inclination - corrected limiting csb of about 24.5@xmath1 adequately sampled by our galaxy sample , the dsl distribution of galaxies with a lower csb may have a longer tail toward large values unless the distribution of disk galaxies as a function of csb rises very rapidly toward faint values . in the case of ( b ) , the face - on limiting csb of our galaxy sample is too faint to set a useful constraint on the faint part of the csb distribution function . the author is grateful to e. e. salpeter and g. l. hoffman for their comments on this paper and contribution to the early part of our ccd surface photometry campaign ; to the editor , g. bothun , and the referee , s. mcgaugh , for a number of comments that helped improving this paper ; to w. freudling for providing us with the galphot software package which is part of the data reduction tools used in this paper ; to m. schmitz for going over the tables ; and to the staff members of palomar observatory for their assistance during the observation . the observations presented in this paper were made at the palomar observatory as part of a continuing collaborative agreement between the california institute of technology and cornell university . this work was supported in part by jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration . allen , r. j. , & shu , f. h. 1979 , , 227 , 67 bothun , g. d. , schombert , j. m. , impey , c. d. , & schneider , s. t. 1990 , , 360 , 427 bothun , g. , impey , c. , and mcgaugh , s. 1997 , , 109 , 745 briggs , f. h. 1997 , , 484 , l29 burstein , d. , haynes , m. p. , & faber , s. m. 1991 , nature , 353 , 515 byun , y .- i . 1993 , , 105 , 993 cornell , m. e. , aaronson , m. , bothun , g. , & mould , j. 1987 , , 64 , 507 dalcanton , j. j. , spergel , d. n. , & summers , f. j. 1997a , , 482 , 659 dalcanton , j. j. , spergel , d. n. , gunn , j. e. , schmidt , m. , & schneider , d. p. 1997b , , 114 , 635 davies , j. i. 1990 , 244 , 8 de jong , r. s. 1996 , , 313 , 45 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , buta , r. j. , paturel , g. , & foueu , p. 1991 , third reference catalogue of bright galaxies ( new york : springer ) ( rc3 ) disney , m. j. 1976 , , 263 , 573 disney , m. j. , & phillipps , s. 1983 , , 205 , 1253 disney , m. j. , davies , j. , & phillipps , s. 1989 , , 239 , 939 freeman , k. c. 1970 , , 160 , 811 giovanelli , r. , haynes , m. p. , salzer , j. j. , wegner , g. , costa , l. n. , freudling , w. 1994 , , 107 , 2036 gunn , j. e. , et al . 1987 , opt . engineering , 26 , 779 hoffman , g. l. , dickey , j. , lu , n. y. , & fromhold - treu , r. 1996 , , 473 , 822 johnson , h. l. , & morgan , w. w. 1953 , , 117 , 113 kent , s. m. 1985 , , 59 , 115 kormendy , j. 1977 , , 217 , 406 landolt , a. u. 1983 , , 88 , 439 lu , n. y. , hoffman , g. l. , groff , t. , roos , t. , & lamphier , c. 1993 , , 88 , 383 ( paper i ) lu , n. y. , salpeter , e. e. , & hoffman , g. l. 1994 , , 426 , 473 malmquist , k. g. 1920 , medd . lunds astron . ii , no . 22 mcgaugh , s. s. 1996 , , 280 , 337 mcgaugh , s. s. , bothun , g. d. , & schombert , j. m. 1995 , , 110 , 573 nilson , p. 1973 , uppsala general catalogue of galaxies , uppsala astron . obs . ann . , 6 ( ugc ) pierce , m. j. , & tully , r. b. 1988 , , 330 , 579 schombert , j. m. , bothun , g. d. , schneider , s. e. , mcgaugh , s. s. 1992 , , 103 , 1107 sprayberry , d. , impey , c. d. , & irwin , m. j. 1996 , , 463 , 535 szomoru , a. , guhathakurta , p. , van gorkom , j. h. , knapen , j. h. , weinberg , d. h. , & fruchter , a. s. 1994 , , 108 , 491 tully , r. b. , & fouqu , p. 1985 , , 58 , 67 tully , r. b. , & verheijen , m. a. w. 1997 , , 484 , 145 tully , r. b. , & verheijen , m. a. w. , pierce , m. j. , huang , j .- s . , wainscoat , r. j. 1996 , , 112 , 2471 valentijn , e. a. 1990 , nature , 346 , 153 valentijn , e. a. 1994 , mnras , 266 , 614 van der kruit , p. c. 1987 , , 173 , 59 xu , c. , & buat , v. 1995 , , 293 , l65 i. the ugc & sample : u11880 & i1420 & 25.33 & 18.7 & 36.4 & 0.306 & 47 & 112 & 14.20 & 1.41 & 7.6 & 0.93 & 19.94 & 14.17 & -17.85 & 1.54 u11921 & & 25.63 & 31.6 & 50.9 & 0.650 & 72 & 124 & 14.62 & 2.09 & 12.5 & 1.55 & 20.50 & 14.56 & -17.48 & 1.37 u11968 & n7241 & 22.71 & 47.6 & 101.9 & 0.637 & 71 & 21 & 13.02 & 3.41 & 17.5 & 1.93 & 19.60 & 12.99 & -18.79 & 1.94 u12035 & n7280 & 27.91 & 33.7 & 65.6 & 0.356 & 51 & 74 & 13.09 & 2.68 & 17.2 & 2.33 & 20.88 & 13.04 & -19.19 & 2.10 u12045 & n7290 & 42.04 & 15.0 & 51.8 & 0.449 & 58 & 161 & 14.03 & 1.83 & 11.1 & 2.26 & 20.58 & 13.99 & -19.13 & 1.58 u12074 & & 29.42 & 11.1 & 23.9 & 0.281 & 45 & 149 & 14.71 & 0.84 & 3.6 & 0.51 & 18.32 & 14.70 & -17.64 & 2.04 u12118 & n7328 & 40.67 & 32.9 & 70.5 & 0.604 & 69 & 86 & 13.95 & 2.38 & 15.3 & 3.02 & 20.90 & 13.90 & -19.15 & 1.96 u12151 & & 25.94 & 34.1 & 80.5 & 0.278 & 44 & 0 & 14.80 & 2.61 & 45.2 & 5.69 & 24.12 & 14.17 & -17.90 & 1.36 u12178 & & 28.54 & 43.3 & 92.7 & 0.453 & 58 & 9 & 13.06 & 4.72 & 45.2 & 6.26 & 22.52 & 12.87 & -19.75 & 1.24 u12270 & n7437 & 31.28 & 39.9 & 70.7 & 0.069 & 21 & 43 & 13.57 & 2.35 & 17.2 & 2.61 & 21.51 & 13.48 & -19.00 & 1.48 u12294 & n7448 & 32.37 & 38.8 & 83.1 & 0.511 & 62 & 171 & 12.24 & 3.24 & 16.2 & 2.54 & 19.41 & 12.22 & -20.33 & 1.48 u12316 & n7463 & 28.02 & 43.7 & 113.2 & 0.688 & 75 & 93 & 13.37 & 3.45 & 27.1 & 3.69 & 21.85 & 13.34 & -18.90 & 1.42 u12315 & n7464 & 34.26 & 17.0 & 30.1 & 0.348 & 50 & 154 & 14.71 & 1.30 & 11.2 & 1.86 & 22.22 & 14.64 & -18.03 & 1.01 u12317 & n7465 & 29.35 & 34.6 & 67.5 & 0.230 & 40 & 165 & 13.30 & 2.31 & 25.9 & 3.68 & 23.03 & 13.20 & -19.14 & 1.72 u12329 & n7468 & 30.89 & 17.9 & 38.3 & 0.274 & 44 & 19 & 14.19 & 1.24 & 9.1 & 1.35 & 21.51 & 14.16 & -18.29 & 1.30 u12343 & n7479 & 34.71 & 83.7 & 122.6 & 0.254 & 42 & 39 & 11.81 & 4.36 & 18.7 & 3.15 & 18.35 & 11.78 & -20.92 & 1.99 u12350 & & 31.66 & 41.6 & 89.1 & 0.733 & 79 & 95 & 14.48 & 3.64 & 33.9 & 5.21 & 22.47 & 14.28 & -18.22 & 1.52 u12392 & n7497 & 25.93 & 59.3 & 127.1 & 0.800 & 90 & 42 & 12.98 & 6.71 & 51.7 & 6.50 & 21.63 & 12.88 & -19.19 & 1.87 u12442 & n7537 & 38.25 & 32.2 & 62.8 & 0.717 & 78 & 77 & 13.85 & 2.68 & 16.7 & 3.10 & 20.72 & 13.81 & -19.10 & 1.81 u12447 & n7541 & 38.23 & 53.2 & 103.6 & 0.676 & 74 & 100 & 12.37 & 4.24 & 21.3 & 3.95 & 19.39 & 12.35 & -20.56 & 2.06 u12529 & n7625 & 24.83 & 25.5 & 54.6 & 0.106 & 27 & 32 & 12.86 & 1.84 & 9.3 & 1.12 & 19.50 & 12.84 & -19.13 & 2.02 u12613 & & & & & & & & & & & & & & & u12682 & & 21.57 & 24.3 & 52.1 & 0.378 & 53 & 43 & 14.35 & 1.88 & 12.9 & 1.35 & 21.25 & 14.27 & -17.40 & 1.27 u12699 & n7714 & 39.59 & 27.0 & 57.9 & 0.350 & 50 & 37 & 13.04 & 2.42 & 19.1 & 2.66 & 21.85 & 12.99 & -20.00 & 1.69 u12702 & n7716 & 36.48 & 26.3 & 62.0 & 0.196 & 37 & 45 & 13.03 & 2.56 & 16.7 & 2.95 & 20.98 & 12.97 & -19.84 & 1.85 u12709 & & 37.88 & 43.6 & 77.2 & 0.369 & 52 & 144 & 14.69 & 2.63 & 31.9 & 5.87 & 23.29 & 14.31 & -18.58 & 1.44 u12710 & & 36.58 & 23.6 & 41.9 & 0.266 & 43 & 162 & 14.48 & 1.57 & 11.3 & 2.01 & 21.44 & 14.40 & -18.42 & 1.20 u12737 & n7731 & 40.81 & 29.7 & 47.8 & 0.210 & 38 & 91 & 14.26 & 1.63 & 10.1 & 2.01 & 20.74 & 14.19 & -18.86 & 1.88 u12738 & n7732 & 41.09 & 30.2 & 58.8 & 0.677 & 74 & 89 & 14.33 & 2.22 & 12.6 & 2.51 & 20.25 & 14.29 & -18.78 & 1.51 u12760 & n7742 & 24.65 & 24.5 & 63.6 & 0.029 & 14 & 50 & 12.50 & 2.13 & 12.6 & 1.51 & 20.47 & 12.47 & -19.49 & 1.87 u12759 & n7743 & 25.40 & 51.8 & 91.8 & 0.231 & 40 & 75 & 12.42 & 3.25 & 18.7 & 2.31 & 20.29 & 12.38 & -19.64 & 2.16 u12777 & n7750 & 41.47 & 32.5 & 57.6 & 0.517 & 63 & 173 & 13.56 & 1.99 & 8.5 & 1.71 & 18.31 & 13.54 & -19.55 & 1.77 u12788 & n7757 & 41.71 & 42.6 & 82.9 & 0.305 & 47 & 123 & 13.09 & 2.92 & 18.4 & 3.72 & 20.75 & 13.03 & -20.07 & 1.26 u12843 & & 26.53 & 34.8 & 74.7 & 0.525 & 63 & 24 & 13.14 & 4.38 & 36.2 & 4.65 & 22.04 & 13.00 & -19.12 & 1.12 u12856 & & 26.45 & 39.2 & 69.5 & 0.696 & 76 & 15 & 14.15 & 3.17 & 26.5 & 3.40 & 22.07 & 14.03 & -18.08 & 0.96 u12885 & n7800 & 25.84 & 42.7 & 91.5 & 0.506 & 62 & 45 & 13.28 & 2.82 & 15.5 & 1.94 & 20.07 & 13.26 & -18.80 & 1.36 u00008 & n7814 & & & & & & & & & & & & & & u00075 & n0014 & 14.19 & 35.4 & 75.8 & 0.361 & 51 & 28 & 13.28 & 2.68 & 18.4 & 1.27 & 21.19 & 13.22 & -17.54 & 1.69 u00099 & & 25.77 & 29.1 & 56.6 & 0.357 & 51 & 154 & 15.00 & 2.08 & 27.1 & 3.39 & 23.46 & 14.65 & -17.41 & 1.24 u00119 & & 29.66 & 15.4 & 36.4 & 0.504 & 62 & 81 & 14.62 & 1.29 & 7.8 & 1.12 & 20.59 & 14.59 & -17.77 & 1.34 u00156 & & 17.60 & 32.7 & 77.2 & 0.574 & 67 & 0 & 14.62 & 2.69 & 25.9 & 2.21 & 22.53 & 14.42 & -16.81 & 1.58 u00167 & n0063 & 17.87 & 25.9 & 50.6 & 0.401 & 54 & 104 & 13.17 & 1.76 & 7.6 & 0.66 & 18.45 & 13.16 & -18.10 & 1.89 u00191 & & 17.61 & 45.2 & 66.2 & 0.376 & 52 & 154 & 14.42 & 2.21 & 14.3 & 1.22 & 20.89 & 14.32 & -16.91 & 1.60 u00231 & n0100 & 13.82 & 43.4 & 123.9 & 0.860 & 90 & 55 & 13.84 & 4.66 & 31.9 & 2.14 & 21.11 & 13.77 & -16.93 & 1.71 u00260 & & 30.76 & 37.8 & 89.1 & 0.843 & 90 & 22 & 14.03 & 3.71 & 22.6 & 3.37 & 20.57 & 13.98 & -18.46 & 1.57 u00313 & & 29.84 & 14.0 & 36.4 & 0.411 & 55 & 13 & 14.62 & 1.21 & 6.7 & 0.97 & 20.11 & 14.59 & -17.78 & 1.61 u00668 & i1613 & & & & & & & & & & & & & & u00685 & & 4.27 & 18.5 & 52.7 & 0.384 & 53 & 116 & 14.32 & 1.95 & 13.6 & 0.28 & 21.30 & 14.23 & -13.92 & 1.51 u00711 & & & & & & & & & & & & & & & u00763 & n0428 & 16.76 & 61.8 & 90.5 & 0.140 & 31 & 73 & 12.17 & 3.57 & 20.5 & 1.66 & 20.31 & 12.12 & -19.00 & 1.43 u00858 & n0470 & 33.04 & 48.0 & 85.0 & 0.446 & 58 & 151 & 12.69 & 2.89 & 12.8 & 2.05 & 18.57 & 12.67 & -19.93 & 1.98 u00859 & n0473 & & & & & & & & & & & & & & u00864 & n0474 & 32.85 & 34.8 & 99.2 & 0.155 & 33 & 10 & 12.62 & 3.24 & 24.1 & 3.84 & 21.59 & 12.56 & -20.02 & 2.35 u00895 & n0485 & 31.56 & 26.5 & 62.4 & 0.699 & 76 & 4 & 14.24 & 1.99 & 11.0 & 1.68 & 20.05 & 14.22 & -18.28 & 1.91 u00907 & n0488 & 31.80 & 61.2 & 119.2 & 0.244 & 41 & 4 & 11.28 & 5.86 & 38.8 & 5.98 & 20.91 & 11.23 & -21.28 & 2.18 u00908 & n0489 & 35.09 & 20.5 & 43.9 & 0.747 & 80 & 120 & 13.73 & 1.68 & 7.3 & 1.25 & 18.51 & 13.72 & -19.01 & 2.30 u00914 & n0493 & 32.46 & 58.1 & 113.3 & 0.805 & 90 & 61 & 12.82 & 5.23 & 29.4 & 4.62 & 20.10 & 12.79 & -19.77 & 1.27 u00947 & n0514 & 34.84 & 53.4 & 114.4 & 0.271 & 44 & 102 & 12.64 & 4.01 & 29.4 & 4.96 & 21.52 & 12.54 & -20.17 & 1.71 u00952 & n0518 & 37.73 & 26.3 & 75.0 & 0.650 & 72 & 97 & 14.37 & 2.37 & 16.5 & 3.01 & 21.29 & 14.32 & -18.56 & 2.48 u00966 & n0520 & & & & & & & & & & & & & & u00970 & n0522 & 37.97 & 40.1 & 86.0 & 0.835 & 90 & 33 & 14.14 & 3.35 & 19.4 & 3.57 & 20.32 & 14.09 & -18.81 & 2.51 u00982 & n0532 & 33.41 & 57.5 & 112.0 & 0.729 & 79 & 29 & 13.66 & 3.65 & 24.1 & 3.91 & 21.04 & 13.60 & -19.02 & 2.44 u01014 & & 29.94 & 22.8 & 44.3 & 0.091 & 25 & 174 & 14.83 & 1.33 & 9.8 & 1.42 & 21.53 & 14.79 & -17.59 & 1.00 u01102 & & 27.42 & 28.0 & 37.2 & 0.184 & 36 & 0 & 14.43 & 1.48 & 8.7 & 1.15 & 20.41 & 14.37 & -17.82 & 0.67 u01104 & & 11.15 & 13.3 & 50.5 & 0.399 & 54 & 6 & 14.49 & 1.45 & 9.4 & 0.51 & 20.96 & 14.49 & -15.75 & 2.31 u01133 & & 27.55 & 32.6 & 63.5 & 0.301 & 46 & 177 & 15.52 & 1.96 & 57.2 & 7.63 & 24.88 & 14.43 & -17.77 & u01149 & n0628 & 10.61 & 81.3 & 158.5 & 0.217 & 39 & 148 & 10.47 & 10.26 & 77.6 & 3.99 & 21.52 & 10.38 & -19.75 & 1.55 u01192 & n0658 & 41.49 & 44.0 & 103.8 & 0.503 & 62 & 28 & 13.27 & 3.42 & 31.0 & 6.24 & 22.41 & 13.17 & -19.92 & 1.71 u01195 & & 12.03 & 64.4 & 125.5 & 0.799 & 88 & 48 & 13.70 & 4.57 & 31.9 & 1.86 & 21.29 & 13.62 & -16.78 & 1.36 u01200 & & 12.45 & 17.5 & 66.5 & 0.451 & 58 & 168 & 14.13 & 1.90 & 11.8 & 0.71 & 20.77 & 14.13 & -16.35 & 1.48 u01201 & n0660 & 13.02 & 79.7 & 128.4 & 0.588 & 68 & 23 & 12.16 & 8.00 & 83.5 & 5.27 & 22.83 & 11.96 & -18.61 & 2.21 u01270 & n0676 & & & & & & & & & & & & & & u01304 & n0693 & 22.11 & 43.2 & 92.7 & 0.597 & 69 & 104 & 13.33 & 3.15 & 27.1 & 2.91 & 22.17 & 13.26 & -18.46 & 2.24 u01356 & n0718 & 24.23 & 35.3 & 91.6 & 0.137 & 30 & 23 & 12.63 & 3.04 & 22.2 & 2.60 & 21.53 & 12.57 & -19.35 & u01463 & n0770 & 34.49 & 22.4 & 39.6 & 0.294 & 46 & 10 & 14.45 & 1.20 & 10.8 & 1.80 & 22.34 & 14.42 & -18.27 & u01466 & n0772 & 34.71 & 75.1 & 133.1 & 0.421 & 56 & 124 & 11.25 & 8.73 & 72.4 & 12.18 & 22.01 & 11.14 & -21.56 & 1.76 ii . fainter & galaxies : z409 - 018 & & 19.93 & 12.6 & 26.9 & 0.479 & 60 & 6 & 15.40 & 0.98 & 6.2 & 0.60 & 20.89 & 15.36 & -16.14 & 1.51 z409 - 040 & & 9.47 & 11.5 & 18.5 & 0.400 & 54 & 135 & 15.67 & 0.71 & 4.8 & 0.22 & 21.19 & 15.63 & -14.25 & 2.33 u00494 & & 27.93 & 15.2 & 29.7 & 0.585 & 68 & 95 & 15.21 & 1.03 & 5.1 & 0.69 & 19.42 & 15.19 & -17.04 & 1.40 u00634 & & 31.38 & 28.4 & 55.2 & 0.567 & 66 & 34 & 15.25 & 2.08 & 20.1 & 3.06 & 22.63 & 15.03 & -17.45 & 1.15 u00871 & & 30.36 & 14.6 & 28.4 & 0.271 & 44 & 133 & 16.40 & 1.00 & 11.7 & 1.72 & 23.20 & 16.10 & -16.31 & 1.61 u00882 & & 32.67 & 24.9 & 44.1 & 0.404 & 55 & 82 & 15.72 & 1.42 & 12.3 & 1.96 & 22.23 & 15.56 & -17.01 & 1.50 z411 - 038 & & 36.84 & 12.4 & 24.1 & 0.229 & 40 & 11 & 15.89 & 0.85 & 7.7 & 1.37 & 22.42 & 15.78 & -17.05 & 1.54 f0120 + 0835 & & 31.26 & 18.0 & 29.0 & 0.015 & 10 & 149 & 15.83 & 0.88 & 6.1 & 0.92 & 21.30 & 15.78 & -16.70 & 1.82 z411 - 042 & & 37.62 & 24.9 & 40.0 & 0.466 & 59 & 153 & 15.33 & 1.21 & 11.7 & 2.13 & 22.58 & 15.27 & -17.61 & 0.74 u00964 & & 38.09 & 19.7 & 31.8 & 0.864 & 80 & 90 & 15.40 & 1.12 & 4.7 & 0.86 & 18.17 & 15.38 & -17.52 & 1.06 f0128 + 0424 & & 28.68 & 10.4 & 24.6 & 0.336 & 49 & 133 & 16.27 & 0.75 & 5.2 & 0.72 & 21.29 & 16.24 & -16.05 & 1.46 ccccc 25 & 90 & 0.33 & 6.1 & 18.9 24 & 32 & 0.58 & 3.9 & 6.7 23 & 20 & 0.58 & 2.4 & 4.2 22 & 13 & 0.57 & 1.5 & 2.7 21 & 8 & 0.60 & 1.0 & 1.6 20 & 6 & 0.50 & 0.6 & 1.3 19 & 5 & 0.37 & 0.4 & 1.1 18 & 4.5 & 0.26 & 0.2 & 1.0	we report new b - band ccd surface photometry on a sample of 76 disk galaxies brighter than @xmath0mag in the uppsala general catalogue of galaxies , which are confined within a volume located in the outer part of the local supercluster . with our earlier published i - band ccd and high s / n - ratio 21 cm hi data ( lu et al . 1993 ) , this paper completes our optical surface photometry campaign on this galaxy sample . as an application of this data set , the b - band photometry is used here to illustrate two selection effects which have been somewhat overlooked in the literature , but which may be important in deriving the distribution function of disk central surface brightness ( csb ) of disk galaxies from a diameter or / and flux limited sample : a malmquist - type bias against disk galaxies with small disk scale lengths ( dsl ) at a given csb ; and a disk inclination dependent selection effect that may , for example , bias toward inclined disks near the threshold of a diameter limited selection if disks are not completely opaque in optical . taking into consideration these selection effects , we present a method of constructing a volume sampling function and a way to interpret the derived distribution function of csb and dsl . application of this method to our galaxy sample implies that if galaxy disks are _ optically thin _ , csb and dsl may well be correlated in the sense that , up to an inclination - corrected limiting csb of about 24.5@xmath1 that is adequately probed by our galaxy sample , the dsl distribution of galaxies with a lower csb may have a longer tail toward large values _ unless _ the distribution of disk galaxies as a function of csb rises rapidly toward faint values .
although we now know that the majority of quasars are , at best , weak radio sources , quasars were first recognized as a result of their radio emission . over the decades a great deal of information has been accumulated about the radio properties of quasars . generally speaking , roughly 10% of quasars are thought to be `` radio - loud '' ( e.g. , * ? ? ? * and references therein ) . the radio emission can be associated with either the quasar itself or with radio lobes many kiloparsecs removed from the quasar ( hereafter we refer these double lobed sources as fr2s ) . traditionally it was widely held that there was a dichotomy between the radio - loud and radio - quiet quasar populations , although more recent radio surveys have cast doubt on that picture ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? the advent of wide area radio surveys like the first survey coupled with large quasar surveys like sdss permit a more extensive inventory of the radio properties of quasars . the association of radio flux with the quasar itself ( hereafter referred to as core emission ) is straightforward given the astrometric accuracy of both the optical and radio positions ( typically better than 1 arcsec ) . the association of radio lobes is more problematic since given the density of radio sources on the sky , random radio sources will sometimes masquerade as associated radio lobes in this paper we attempt to quantify both the core and fr2 radio emission associated with a large sample of optically selected quasars . our new implementation of matching the first radio _ environment _ to its associated quasar goes beyond the simple one - to - one matching ( within a certain small radius , typically 2 ) , in that it investigates ( and ranks ) all the possible radio source configurations near the quasar . this also goes beyond other attempts to account for double lobed radio sources without a detected radio core , most notably by @xcite who matched mid - points between close pairs of radio components to the sdss early data release catalog . while this does recover most ( if not all ) of the fr2 systems that are perfectly straight , it misses sources that are bent . even slight bends in large systems will offset the midpoint enough from the quasar position as to be a miss . the paper is organized as follows . the first few sections ( [ intro1 ] through [ intro2 ] ) describe the matching process of the radio and quasar samples . the results ( [ results ] ) are separated in two parts : one based on statistical inferences of the sample as a whole , and one based on an actual sample of fr2 sources . these two are not necessarily the same . the former section ( [ sampres1 ] through [ redshiftdeps ] ) mainly deals with occurrence levels of fr2 s among quasars , the distribution of core components among these fr2 quasars , and their redshift dependencies . all these results are based on the detailed comparison between the actual and random samples . in other words , it will tell us _ how many _ fr2 quasars there are among the total , however , it does not tell us _ which ones _ are fr2 . this is addressed in the second part of [ results ] , which deals with an _ actual _ sample of fr2 quasars ( see [ samplespecific ] on how we select these ) . this sample forms a representative subsample of the total number of fr2 quasars we infer to be present in the initial sample , and is used to construct an optical composite spectrum of fr2 quasars . section [ compspectra ] details the results of the comparison to radio quiet and non - fr2 radio loud quasar spectra . our quasar sample is based on the sloan digital sky survey ( sdss ) data release 3 ( dr3 , * ? ? ? * ) quasar list , as far as it overlaps with the first survey @xcite . this resulted in a sample of 44984 quasars . in this paper we focus on the radio population properties of optically selected quasars . the radio matching question is not a straightforward one . by just matching the core positions , we are biasing against the fraction of radio quasars which have weak , undetected , cores . therefore , this section is separated in two parts , core matching , and environment matching . the former is the straight quasar - radio positional match to within a fixed radius ( 3 in our case ) , whereas the latter actually takes the distribution of radio sources in the direct vicinity of the quasar in account . this allows us to fully account for the double lobed fr2 type quasars , whether they have detectable cores or not . in this section , we quantify the fraction of quasars that exhibit core emission . we can actually go slightly deeper than the official first catalog , with its nominal @xmath2 lower threshold of 1.0 mjy , by creating @xmath3 lists based on the radio images and quasar positions . this allows us to go down to a detection limit of @xmath4 mjy ( versus 1.0 mjy for the official version ) . given the steeply rising number density distribution of radio sources toward lower flux levels , one might be concerned about the potential for an increase in false detections at sub - mjy flux density levels . the relative optical to radio astrometry is , however , accurate enough to match within small apertures ( to better than 3 ) , reducing the occurrence of chance superpositions significantly . the surface density of radio sources at the 1 mjy level is not high enough to significantly contaminate the counts based on 3 arcsecond matching . the fraction of radio core detected quasars ( rcdq ) out of the total quasar population hovers around the 10% mark , but this is a strong function of the radio flux density limit . it also depends on the initial selection of the quasar sample . the sdss quasar selection is mainly done on their optical colors @xcite , but @xmath5 of the quasars were selected solely on the basis of their radio emission ( 1397 out of 44984 ) . looking at only those sdss quasars which have been selected expressly on their optical colors ( see the quasar selection flow - chart of richards et al . 2002 , fig 1 ) , there are 34147 sources which have either a qso_cap , qso_skirt , and / or a qso_hiz flag set . for these , the relevant radio core detection fractions are 7.1% ( 2430 ) and 10.1% ( 3458 ) for the 5@xmath6 and 3@xmath6 detection limits , respectively ( the binomial error on these percentages is on the order of 0.1% ) . these core fractions are higher for the 10837 quasars ( 44984@xmath734147 ) that made it into the sdss sample via other means ( 1694 , 15.6% and 1855 , 17.1% for the 5 and 3@xmath6 catalogs ) . the higher core fractions are due to the large number of targeted first sources that would not have made it into the sample otherwise , and to the greater likelihood of radio emission among x - ray selected quasars . clearly , the initial quasar selection criteria impact the rate at which their cores are detected by first . the results have been summarized in table 1 . a more direct view of the flux limit dependence of the rcdq fraction is offered by fig . [ corematches ] . an extrapolation of the data suggests that at 0.1 mjy about 20% of quasar cores will be detected . this extrapolation is not unrealistic , and even may be an underestimate : the extragalactic spitzer first look survey ( fls ) field has been covered by both the sdss and the vla down to 0.1 mjy levels @xcite . out of the 33 quasars in the dr3 that are covered by this vla survey , we recover 12 using the exact same matching criterion . this corresponds to a fraction of 36% , which carries an 8% formal @xmath8 uncertainty . in fact , judging by the progression of detection rate in fig . [ corematches ] , one does not have to go much deeper than 0.1 mjy to recover the majority of ( optically ) selected quasars . the results and discussion presented in this paper , however , are only relevant to the subset of quasars with cores brighter than @xmath9 mjy . it is this @xmath10 of the total that is well - sampled by the first catalog . this should be kept in mind as well for the sections where we discuss radio quasar morphology . the first catalog is essentially a catalog of components , and not a list of sources . this means that sources which have discrete components , like the fr2 sources we are interested in , are going to have multiple entries in the first catalog . if one uses a positional matching scheme as described in the last section , and then either visually or in an automated way assesses the quasar morphology , one will find a mix of core- and lobe - dominated quasars _ provided _ that the core has been detected . however , this mechanism is going to miss the fr2 sources without a detected core , thereby skewing the quasar radio population toward the core dominated sources . preferably one would like to develop an objective procedure for picking out candidate fr2 morphologies . we decided upon a catalog - based approach where the first catalog was used to find all sources within a 450 of a quasar position ( excluding the core emission itself ) . sources around the quasar position were then considered pairwise , where each pair was considered a potential set of radio lobes . pairs were ranked by their likelihood of forming an fr2 based on their distances to the quasar and their opening angle as seen from the quasar position . higher scores were given to opening angles closer to 180 degrees , and to smaller distances from the quasar . the most important factor turned out to be the opening angle . nearby pairs of sources unrelated to the quasar will tend to have small opening angles as will a pair of sources within the same radio lobe of the quasar , so we weighted against candidate fr2 sources with opening angles smaller than 50 . the chances of these sources to be real are very small , and even if they are a single source , their relevance to fr2 sources will be questionable . we score the possible configurations as follows : @xmath11 where @xmath12 is the opening angle ( in degrees ) , and @xmath13 and @xmath14 are the distance rank numbers of the components under consideration . the closest component to the quasar has an @xmath15 , the next closest is @xmath16 , etcetera . this way , the program will give the most weight to the radio components closest to the quasar , irrespective of what that separation turns out to be in physical terms . each quasar which has at least 2 radio components within the 450search radius will get an assigned `` most likely '' fr2 configuration ( i.e. , the configuration with the highest score @xmath17 . this , by itself , does not mean it is a real fr2 . in fact , this procedure turns up large numbers of false positives . therefore , as a control , we searched for fr2 morphologies around a large sample of random sky positions that fall within the area covered by first . since all of the results for fr2s depend critically on the quality of the control sample , we increased this random sample size 20-fold over the actual quasar sample ( of 44984 ) . given the area of the first survey ( @xmath18 sq . degree ) and our matching area ( 1/20th of a sq . degree ) , a much larger number of pointings would start to overlap itself too much ( i.e. , the random samples would not be independent of each other ) . in fig . [ pahist ] we display a set of histograms for particular fr2 sizes . for each , the number fr2 candidates are plotted as a function of opening angle both around the true quasar position ( light - grey trace ) as well as the offset positions ( dark - grey trace ) . there is a clear excess of nominal fr2 sources surrounding quasar positions which we take as a true measure of the number of quasars with associated fr2s . although the distribution of fr2s has a pronounced peak at opening angles of 180 degrees , the distribution is quite broad , extending out to nearly 90 degrees . it is possible that some of this signal results from quasars living within ( radio ) clusters and hence being surrounded by an excess of unrelated sources , but such a signal should not show a strong preference for opening angles near 180 degrees . the set of histograms also illustrates the relative importance of chance fr2 occurrences ( dark - grey histograms ) , which become progressively more prevalent if one starts considering the larger fr2 configurations . while the smallest size bin does have some contamination ( @xmath19 on average across all opening angles ) , almost all of the signal beyond opening angles of 90 degrees is real ( less than 5% contamination for these angles ) . however , the significance of the fr2 matches drops significantly for the larger sources . more than 92% of the signal in the 3 to 4 arcminute bin is by chance . clearly , most of the suggested fr2 configurations are spurious at larger diameter , and only deeper observations and individual inspection of a candidate source can provide any certainty . in the next few sections we describe the results of the analysis . the primary result we can quantify is the fraction of quasars that can be associated with a double lobed radio structure ( whether a core has been detected in the radio or not ) . this is different from the discussion in [ faintcorematches ] which relates to the fraction of quasars that have radio emission at the quasar core position . this value , while considerably higher than the rates for the fr2 quasars , does not form an upper limit to the fraction of quasars associated with radio emission : some of the fr2 quasars do not have a detected radio core . figure [ pahist ] depicts the excess number of fr2 quasars over the baseline values , plotted for progressively larger radio sources . the contamination rates go up as more random first components fall within the covered area , and , at the same time , fewer real fr2 sources are found . this effect is illustrated in fig . [ candexcess ] , which shows the fr2 excesses as function of overall source size . the light - grey line indicates the fr2 number counts for candidates _ without _ a detected ( 3@xmath6 ) core , and the dark - grey histogram is for the fr2 candidates _ with _ a detected core . it is clear that fr2 sources larger than about 300 are very rare , and basically can not be identified using this method . most fr2 sources are small , with the bulk being having diameters of less than 100 . the summed total excess numbers , based on fig . [ candexcess ] and limited to 300 or smaller , are 749 fr2 candidates ( 1.7% of the total ) , of which 547 have cores . some uncertainties in the exact numbers still remain , particularly due to the noise in the counts at larger source sizes . a typical uncertainty of @xmath20 should be considered on these numbers ( based on variations in the fr2 total using different instances of the random position catalog ) . at these levels , it is clear that the fr2 phenomenon is much less common than quasar core emission ; 1.7% versus 10% ( see [ faintcorematches ] ) . indeed , of all the quasars with a detected radio core , only about 1 in 9 is also an fr2 . the relative numbers have been recapitulated in table 2 . as noted above , not all fr2 quasars have cores that are detected by first . we estimate that about 75% of fr2 sources have detected cores down to the 0.75 mjy flux density level . this value compares well with the number for our `` actual '' fr2 quasar sample of [ samplespecific ] . out of 422 fr2 quasar sources , 265 have detected cores ( 62.8% ) down to the first detection limit ( 1 mjy ) . we are now in the position of investigating whether there is a correlation between the overall size of the fr2 and the presence of a radio core . in orientation dependent unification schemes , a radio source observed close to the radio jet axis will be both significantly foreshortened and its core brightness will be enhanced by beaming effects ( e.g. , barthel 1989 , hoekstra et al . this would imply that , given a particular distribution of fr2 radio source sizes and core luminosities , the smaller fr2 sources would be associated ( on average ) with brighter core components . this should translate into a higher fraction of detected cores among smaller fr2 quasars ( everything else being equal ) . figure [ corefrac ] shows the fraction of fr2 candidates that have detected cores , as function of overall size . there does not appear to be a significant trend toward lower core - fractions as one considers larger sources . the much lower fraction for the very smallest size bin is due to the limited resolution of the first survey ( about 5 ) , which makes it hard to isolate the core from the lobe emission for sources with an overall size less than about half an arcminute . also , beyond about 275 the core - ratio becomes rather hard to measure ; not a lot of fr2 candidates are this large ( see fig . [ candexcess ] ) . since the core - fraction is more or less constant , and does not depend on the source diameter , it does not appear that relativistic beaming is affecting the ( faint ) core counts . unfortunately , one expects the strongest core beaming contributions for the smallest sources ; exactly the ones that are most affected by our limited resolution . the angular distributions in fig . [ pahist ] reveal a large number of more or less _ bent _ fr2 sources . bends in fr2 sources can be due to a variety of mechanisms , either intrinsic or extrinsic to the host galaxy . local density gradients in the host system can account for bending ( e.g. , * ? ? ? * ) , or radio jets can run into overdensities in the ambient medium , resulting in disruption / deflection of the radio structure ( e.g. , * ? ? ? extrinsic bending of the radio source can be achieved through interactions with a ( hot ) intracluster medium . any space motion of the source through this medium will result in ram - pressure bending of the radio structure ( e.g. , * ? ? ? * ; * ? ? ? * ) . and finally , radio morphologies can be severely deformed by merger events ( e.g. , * ? ? ? regardless of the possible individual mechanisms , a large fraction of our fr2 quasars have significant bending : only slightly more than 56% of fr2 quasars smaller than 3 arcminutes have opening angles larger than 170 degrees ( this value is 65% for the actual sample of [ samplespecific ] ) . this large fraction of bent quasars is in agreement with earlier findings ( based on smaller quasar samples ) of , e.g. , @xcite . we can investigate whether there are trends with quasar redshift based on statistical arguments . this is done by subdividing the sample of 44984 in two parts ( high and low redshift ) , and then comparing the results for each subsample . as with the main sample , each subsample has its own control sample providing the accurate baseline . previous studies ( e.g. , blundell et al . 1999 ) have suggested that fr2 sources appear to be physically smaller at larger redshifts . for self - similar expansion the size of a radio source relates directly to its age . it also correlates with its luminosity , since , based on relative number densities of symmetric double lobed sources over a large range of sizes ( e.g. , fanti et al . 1995 , odea & baum 1997 ) , one expects a significant decline in radio flux as a lobe expands adiabatically . while the picture in a fixed flux density limit will preferentially bias against older ( and therefore fainter ) radio sources at higher redshifts , resulting in a `` youth - size - redshift '' degeneracy , scant hard evidence is available in the literature . indeed , several studies contradict each other . see blundell et al . ( 1999 ) for a nice summary . we , however , are in a good position to address this issue . first , our quasar sample has not been selected from the radio , and as such has perhaps less radio bias built in . also , we have complete redshift information on our sample . the redshift range is furthermore much larger than in any of the previous studies . the median redshift for our sample of quasars is 1.3960 , which results in mean redshifts for the low- and high - redshift halves of the sample of @xmath21 , and @xmath22 . the first test we can perform on the 2 subsamples is to check whether their relative numbers as function of the average lobe flux density makes sense . the results are plotted in fig . [ zdep ] , left panel . the curves are depicting the cumulative number of fr2 candidates _ smaller than 100 _ for which the mean lobe flux density is larger than a certain value . we have explicitly removed any core contribution to this mean . we limited our comparison here to the smaller sources , for which the background contamination is smallest . since the left panel shows the results in the observed frame , it is clear that we detect far more local fr2 sources than high redshift ones ( light - grey curve in comparison to the dark - grey one ) . on average , more than twice as many candidates fall in the low - redshift bin compared to the high - redshift ones ( 408 candidates versus 201 ) . furthermore , the offset between the two curves appears to be fairly constant , indicating that the underlying population properties may not be that different ( i.e. , we can match both by shifting the high - redshift curve along the x - axis , thereby correcting for the lower lobe flux densities due to their larger distances ) . this is exactly what we have done in the right panel . all of the fr2 candidates have been put at a fiducial redshift of 1 , correcting its lobe emission both for relative distances and intrinsic radio spectral index . for the former we assumed a wmap cosmology ( h@xmath23 km s@xmath24 mpc@xmath24 , @xmath25 , and @xmath26 ) , and for the latter we adopted @xmath27 for the high frequency part of the radio spectrum ( @xmath28 ghz ) . it should be noted that these cumulative curves are only weakly dependent on the cosmological parameters @xmath29 and radio spectral index @xmath30 . it is not dependent on the hubble constant . the physical maximum size for the fr2 candidates is set at 800kpc , which roughly corresponds to the 100 size limit in the left panel . both curves agree reasonably well now . at the faint end of each curve , incompleteness of the first survey flattens the distribution . this accounts for the count mismatch between the light- and dark - grey curves below about 10 mjy . on the other end of both curves , low number statistics increase the uncertainties . the slightly larger number of bright fr2 sources for the high - redshift bin ( see fig . [ zdep ] , right panel ) may be real ( i.e. , fr2 sources are brighter at high redshifts compared to their low - redshift counterparts ) , but the offset is not significant . also note the effect of changing the radio spectral index from @xmath31 to @xmath32 ( dotted dark - grey line versus solid line ) . a negative @xmath30 value has the effect of increasing the lobe fluxes , especially for the higher redshift sources . flattening the @xmath30 to @xmath32 ( or toward even more unrealistic positive values ) therefore acts to lower the average lobe fluxes , and as a consequence both cumulative distributions start to agree better . this would also suggest that the high - redshift sources may be intrinsically brighter , and that only by artificially lowering the fluxes can both distributions made to agree . this brings us to the second question regarding redshift dependencies : are the high - redshift fr2 quasars intrinsically smaller because we are biased against observing older , fainter , and larger radio sources ? to this end we used the same two datasets that were used for fig . [ zdep ] , right panel . the upper size limit is set at 800kpc for both subsets , but this does not really affect each size distribution , since there are not that many fr2 sources this large . figure [ histsizes ] shows both distributions for the low and high redshift bins ( same coloring as before ) . as in fig . [ corefrac ] , the smallest size bins are affected by resolution effects , albeit it is easier to measure the lobe separation than whether or not there is a core component between the lobes . the smallest fr2 sources in our sample are about 10 ( @xmath3380 kpc at @xmath34 ) , which is a bit better than the smallest fr2 for which a clear core component can be detected ( @xmath35 ) . the apparent peak in our size distributions ( around 200 kpc ) agrees with values found for 3cr quasars . @xcite quote a value of @xmath36 kpc , though given the shape of the distribution it is not clear how useful this measure is . a kolmogorov - smirnoff test deems the difference between the low - redshift ( green histogram ) and high - redshift ( blue histogram ) size distributions insignificant . therefore , it does not appear that there is evidence for fr2 sources to be smaller in the earlier universe . an issue that has been ignored so far is that we tacitly assumed that the fr2 sizes for these quasars are accurate . if these sources have orientations preferentially toward our line of sight ( and we are dealing with quasars here ) , significant foreshortening may underestimate their real sizes by quite a bit ( see antonucci 1993 ) . this will also `` squash '' both distributions toward the smaller sizes , making it hard to differentiate the two . previous studies ( e.g. , * ? ? ? * ) relied on ( smaller ) samples of radio galaxies , for which the assumption that they are oriented in the plane of the sky is less problematic . other studies which mainly focused on fr2 quasars ( e.g , nilsson et al . 1998 , teerikorpi 2001 ) also do not find a size - redshift correlation . the next few sections deal with properties _ intrinsic _ to fr2 quasars . as such , we need a subsample of our quasar list that we feel consists of genuine fr2 sources . we know that out of the total sample of 44984 about 750 are fr2 sources , however , we do not know which ones . what we can do is create a subsample that is guaranteed to have less than 5% of non - fr2 source contamination . this is done by stepping through the multidimensional space spanned up by histograms of the fig . [ pahist ] type as function of overall size . as can be seen in fig . [ pahist ] , the bins with large opening angles only have a small contamination fraction ( in this case for sources smaller than 100 ) . obviously , the signal - to - noise goes down quite a bit for larger overall sizes , and progressively fewer of those candidates are real fr2 . by assembling all the quasars in those bins that have a contamination rate less than 5% , as function of opening angle and overall size , we constructed a fr2 quasar sample that is more reliable than 95% . it contains 422 sources , and forms the basis for our subsequent studies . sample properties are listed in table 3 , and the positions of the 417 actual fr2 quasars are given in table 4 . the different radio morphological properties of the fr2 sources have been used with varying degrees of success to infer its physical properties . in particular , these are : the observed asymmetries in the arm - length ratio ( @xmath37 , here defined as the ratio of the larger lobe - core to the smaller lobe - core separation ) , the lobe flux density ratio ( @xmath38 ) , and the distribution of the lobe opening angle ( @xmath12 , with a linear source having a @xmath12 of 180 ) . @xcite provide a nice historic overview of the literature on these parameters . as can be inferred from a median flux ratio value of @xmath39 ( see table 5 ) , the closer lobe is also the brightest . this is consistent with the much earlier findings of @xcite for the 3cr catalog , and implies directly that the lobe advance speeds are not relativistic , and that most of the arm - length and flux density asymmetries are intrinsic to the source ( and not due to orientation , relativistic motions , and doppler boosting ) . if we separate the low and high - redshift parts of our sample , we can test whether any trend with redshift appears . @xcite , for instance , suggested that quasars are more bent at high redshifts . in our sample we do not find a strong redshift dependency . the median opening angles are 173.6 and 172.7 , for the low and high redshift bins respectively ) , nor is it dependent on redshift . ] . a kolmogorov - smirnoff test deemed the two distributions different at the 97.2% confidence level ( a 2.2@xmath6 results ) . this would marginally confirm the barthel & miley claim . however , @xcite quote a 2@xmath6 result in the opposite sense , albeit using a much smaller sample ( 23 quasars ) . we also found no significant differences between the low and high - redshift values of the arm - length ratios @xmath37 ( ks - results : different at the 87.0% level , 1.51@xmath6 ) , and the flux ratios @xmath40 ( similar at the 97.0% level , 2.2@xmath6 ) . the mackay - type asymmetry , in which the nearest lobe is also the brightest , is not found to break down for the brightest of our quasars . if we separate our sample into a low- and high - flux bin ( which includes the core contribution ) , we do not see a reversal in the flux asymmetry toward the most radio luminous fr2 sources ( e.g. , * ? ? ? * and references therein ) . actually , for our sample we find a significant ( 3.25@xmath6 ) trend for the brightest quasars to adhere more to the mackay - asymmetry than the fainter ones . using the same matching technique as described in the previous section , we made two additional control samples . whereas our fr2 sample is selected based on a combination of large opening angle ( @xmath41 degrees ) and small overall size ( @xmath42 ) , our control samples form the other extreme . very few , if any , genuine fr2 sources will be characterized by radio structures with small opening angles ( @xmath43 degrees ) and large sizes ( @xmath44 ) . therefore , we use these criteria to select two _ non - fr2 _ control samples : one that has a first source coincident with the quasar ( remember that the matching algorithm explicitly excludes components within 3 of the quasar position ) , and another one without a first counterpart to the quasar . for all practical purposes , we can consider the former sample to be quasars which are associated with just one first component ( the `` core dominated '' sample - cd ) , and the latter as quasars without any detected first counterpart ( the `` radio quiet '' sample - rq ) . both of the cd and rq samples initially contained more candidates than the fr2 sample . this allows for small adjustments in the mean sample properties , in particular the redshift distribution . we therefore matched the redshift distribution of the cd and rq samples to the one of the fr2 sample . this resulted in a cd sample which matches the fr2 in redshift - space and in absolute number . the rq sample , which will function as a baseline to both the fr2 and cd samples , contains a much larger number ( 6330 entries ) , but again with an identical redshift distribution . the mean properties of the samples are listed in table 3 . one of the very useful aspects of our sdss based quasar sample is the availability of a large set of complementary data , including the optical spectra for all quasars . an otherwise almost impossible stand - alone observing project due to the combination of low fr2 quasar incidence rates and large datasets , is sidestepped by using the rich sdss data archive . we can therefore readily construct composite optical spectra for our 3 samples ( as listed in table 3 ) . we basically used the same method for the construction of the composite spectrum as outlined by @xcite , in combination with a relative normalization scheme similar to the ones used in @xcite . each composite has been normalized by its continuum flux at 4000 ( restframe ) . the resulting spectra are plotted in fig . [ compview ] , color coded green for the radio - quiet ( rq ) quasar control sample , blue for the core - dominated ( cd ) radio - loud quasars , and red for the lobe dominated ( fr2 ) radio quasars . all three composite spectra are similar to each other and to the composite of @xcite . figure [ nrspec ] shows the number of quasars from each subsample that were used in constructing the composite spectrum . since each individual spectrum has to be corrected by a @xmath45 factor to bring it to its restframe , not all quasars contribute to the same part of the composite . in fact , the quasars that contribute to the shortest wavelengths are not the same that go into the longer wavelength part . this should be kept in mind if one wants to compare the various emission lines . any dependence of the emission line properties on redshift will therefore affect the short wavelength part of the composite more than the long wavelength part ( which is made up of low redshift sources ) . @xcite investigated the effect of dust - absorption on composite quasar spectra ( regardless of whether they are associated with radio sources ) , and we have indicated two of the absorbed template spectra ( composite numbers 5 and 6 , see their fig . 7 ) in our fig . [ compview ] as the red and gray dashed lines , respectively . from this it is clear that our 3 sub - samples do not appear to have significant intrinsic dust - absorption associated with them . indeed , the range of relative fluxes toward the blue end of the spectrum falls within the range of `` normal '' quasars ( templates 1@xmath74 of richards et al . ( 2003 ) ) . the differences in spectral slopes among our 3 samples are real . we measure continuum slopes ( over the range 1450 to 4040 , identical to richards et al . ( 2003 ) ) of : @xmath46 , @xmath47 , and @xmath48 for the fr2 , rq , and cd samples respectively . these values are significantly different from the reddened templates ( @xmath49 and @xmath50 for the red and grey dashed lines in fig . [ compview ] ) , suggesting that our quasars are intrinsically different from dust - reddened quasars ( e.g. , * ? ? ? * ; * ? ? ? in order to study differences in line emission , small differences in spectral slope have to be removed . this is achieved by first normalizing each spectrum to the continuum flux just shortward of the emission line in question . then , by fitting a powerlaw to the local continuum , each emission line spectrum can be `` rectified '' to a slope of unity ( i.e. , making sure both the left and right sides of the zoomed - in spectrum are set to unity ) . a similar approach has been employed by ( * ? ? ? * see their figs . 8 and 9 ) . the results are plotted in fig . [ specslide ] , zoomed in around prominent emission lines . the panels are arranged in order of increasing restframe wavelength . a few key observations can be made . the first , and most striking one , is that fr2 quasars tend to have stronger moderate - to - high ionization emission lines in their spectrum than either the cd and rq samples . this can be seen especially for the , [ ] , [ ] , and [ ] emission lines . the inverse appears to be the case for the balmer lines : the fr2 sources have significantly fainter balmer lines than either the cd or rq samples . notice , for instance , the h@xmath51 , h@xmath52 , h@xmath53 , and h@xmath30 sequence in fig . [ specslide ] . other lines , like and ] , do not seem to differ among our 3 samples . measured line widths , line centers and fluxes for the most prominent emission lines are listed in table 6 . since a lot of the lines have shapes that are quite different from the gaussian form , we have fitted the profiles with the more general form @xmath54 , with @xmath55 a normalization constant , and @xmath56 a free parameter . note that for a gaussian , @xmath57 . the fwhm of the profile can be obtained directly from the values of @xmath56 and @xmath6 : @xmath58 . allowing values of @xmath59 results in better fits for lines with broad wings ( e.g. , @xmath60 $ ] in fig . [ specslide ] ) . typically the difference in equivalent width ( ew ) as fitted by the function and the actual measured value is less than 1% . the fluxes in table 6 have been derived from the measured ew values , multiplied by the continuum level at the center of the line ( as determined by a powerlaw fit , see fig . [ compview ] ) . since all composite spectra have been normalized to a fiducial value of 1.00 at 4000 the fluxes are relative to this 4000 continuum value , and can be compared across the three samples ( columns 6 and 11 in table 6 ) . in addition , we have normalized these fluxes by the value of the ly@xmath30 flux for each subsample . this effectively takes out the slight spectral slope dependency , and allows for an easier comparison to the values of ( * ? ? ? * their table 2 ) . the differences between the various species of emission lines among the 3 subsamples , as illustrated in fig . [ specslide ] , are corroborated by their line fluxes and ratios . even though we can not use emission line ratios ( like @xmath61@xmath60 $ ] / h@xmath30 , see osterbrock 1989 ) to determine whether we are dealing with agn or -region dominated emission regimes ( due to the fact that the broad and narrow lines do not originate from the same region ) , we can still discern trends between the subsamples in the relative importance of broad vs. narrow line emission . this is illustrated by fig . [ linediag ] , in which we have plotted various ratios of narrow and broad emission lines ( based on fluxes listed in table 6 ) . the narrow lines are normalized on the @xmath62-axis by the broad - line h@xmath30 flux , and on the @xmath63-axis by the broad - line component ( listed separately in table 6 ) of the h@xmath53 line . it is clear from this plot that , as one progresses from rq , cd , to fr2 sample , the relative importance of various narrow lines increases . the offset between the rq and vanden berk samples ( which in principle should coincide ) is in part due to the presence of a narrow - line component in their h@xmath53 fluxes ( lowering the points along the @xmath63-axis ) , and a slightly larger flux density in their composite h@xmath30 line ( moving the points to the left along the @xmath62-axis ) . the offset probably serves best to illustrate the inherent uncertainties in plots like these . so , in summary , it appears that the fr2 sources tend to have brighter moderate - to - high ionization lines , while at the same time having much less prominent balmer lines , than either the cd or rq samples . the latter two have far stronger comparable emission line profiles / fluxes , with the possible exception of the higher balmer lines and [ ] . radio sources are known to interact with their ambient media , especially in the earlier stages of radio source evolution where the structure is confined to within the host galaxy . in these compact stages , copious amounts of line - emission are induced at the interfaces of the radio plasma and ambient medium ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? other types of radio activity related spectral signatures are enhanced star - formation induced by the powerful radio jet ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , scattered nuclear uv light off the wall of the area `` excavated '' by the radio structure ( e.g. , * ? ? ? * ; * ? ? ? * ) , or more generally , direct photo - ionization of the ambient gas by the agn along radiation cones coinciding with the radio symmetry axis ( e.g. , * ? ? ? the last three scenarios are more long - lived ( i.e. , the resulting stars will be around for a while ) , whereas the shock - ionization of the line emission gas is an in - situ event , and will only last as long as the radio source is there to shock the gas ( @xmath64 years ) . it therefore appears reasonable to guess that in the case of the fr2 quasars , such an ongoing interaction between the radio structure and its ambient medium is producing the excess flux in the narrow lines . indeed , shock precursor clouds are found to be particularly bright in high - ionization lines like @xmath61@xmath60 $ ] compared to h@xmath30 ( e.g. , * ? ? ? * ; * ? ? ? since the optical spectrum is taken _ at _ the quasar position , and not at the radio _ lobe _ position , we are obviously dealing with interactions between the gaseous medium and the radio core ( whether we detected one or not ) . the other sample of quasars associated with radio activity , the core - dominated ( cd ) sample , has optical spectral properties which do not differ significantly from radio - quiet quasars . we have combined a sample of 44984 sdss quasars with the first radio survey . instead of comparing optical and radio positions for the quasars directly to within a small radius ( say , 3 ) , we matched the quasar position to its complete radio _ environment _ within 450 . this way , we are able to characterize the radio morphological make - up of what is essentially an optically selected quasar sample , regardless of whether the quasar ( nucleus ) itself has been detected in the radio . the results can be separated into ones that pertain to the quasar population as a whole , and those that only concern fr2 sources . for the former category we list : 1 ) only a small fraction of the quasars have radio emission associated with the core itself ( @xmath65 at the 0.75 mjy level ) ; 2 ) fr2 quasars are even rarer , only 1.7% of the general population is associated with a double lobed radio source ; 3 ) of these , about three - quarter have a detected core ; 4 ) roughly half of the fr2 quasars have bends larger than 20 degrees from linear , indicating either interactions of the radio plasma with the icm or igm ; and 5 ) no evidence for correlations with redshift among our fr2 quasars was found : radio lobe flux densities and radio source diameters of the quasars have similar distributions at low and high redshifts . to investigate more detailed source related properties , we used an actual sample of 422 fr2 quasars and two comparison samples of radio quiet and non - fr2 radio loud quasars . these three samples are matched in their redshift distributions , and for each we constructed an optical composite spectrum using sdss spectroscopic data . based on these spectra we conclude that the fr2 quasars have stronger high - ionization emission lines compared to both the radio quiet and non - fr2 radio loud sources . this may be due to higher levels of shock ionization of the ambient gas , as induced by the expanding radio source in fr2 quasars . we like to thank the referee for comments that helped improve the paper . wdvs work was performed under the auspices of the u.s . department of energy , national nuclear security administration by the university of california , lawrence livermore national laboratory under contract no . the authors also acknowledge support from the national radio astronomy observatory , the national science foundations ( grant ast 00 - 98355 ) , the space telescope science institute , and microsoft . lrr [ relmakeup ] quasars with detected cores ( 3@xmath6 limit ) & 5313 & 11.8% + non - fr2 quasars with detected cores & 4766 & 10.6% + fr2 quasars & 749 & 1.7% + fr2 quasars with detected core & 547 & 73% + fr2 quasars without core & 202 & 27% + lrrrr [ sampleprop ] sample size & 422 & 422 & 6330 + @xmath66-band & [email protected] & [email protected] & [email protected] + @xmath68-band & [email protected] & [email protected] & [email protected] + @xmath69-band & [email protected] & [email protected] & [email protected] + @xmath70-band & [email protected] & [email protected] & [email protected] + @xmath71-band & [email protected] & [email protected] & [email protected] + redshift & [email protected] & [email protected] & [email protected] + lllll [ actsample ] 00 03 45.23 @xmath711 08 18.7 & 00 11 38.44 @xmath710 44 58.2 & 00 43 23.43 @xmath700 15 52.6 & 00 44 13.73 00 51 41.0 & 00 51 15.12 @xmath709 02 08.5 + 00 55 08.55 @xmath710 52 06.2 & 00 59 29.12 @xmath710 17 35.0 & 01 03 29.43 00 40 55.1 & 01 17 49.91 @xmath709 05 54.6 & 01 48 47.61 @xmath708 19 36.3 + 02 35 30.71 @xmath707 05 04.5 & 02 40 55.56 00 12 01.2 & 02 48 02.56 @xmath707 19 15.8 & 02 50 48.67 00 02 07.6 & 02 55 12.17 @xmath700 52 24.5 + 03 03 35.76 00 41 45.0 & 03 04 58.97 00 02 35.7 & 03 12 26.12 @xmath700 37 08.9 & 03 13 18.66 00 36 23.8 & 07 34 28.88 32 50 58.4 + 07 41 00.60 39 55 59.5 & 07 41 25.22 33 33 20.1 & 07 42 42.19 37 44 02.1 & 07 44 51.37 29 20 06.0 & 07 45 41.67 31 42 56.7 + 07 48 32.24 26 34 38.9 & 07 49 37.69 38 04 43.4 & 07 52 05.91 28 02 10.8 & 07 52 06.74 24 51 18.8 & 07 52 28.56 37 50 53.6 + 07 54 04.24 42 58 04.3 & 07 55 37.03 25 42 39.0 & 07 56 25.82 25 09 21.6 & 07 56 43.10 31 02 48.8 & 07 59 58.76 25 04 38.4 + 08 00 52.64 41 57 38.8 & 08 01 29.58 46 26 22.8 & 08 01 39.94 31 53 29.8 & 08 02 20.52 30 35 43.0 & 08 02 50.86 33 35 17.8 + 08 05 55.67 34 41 32.3 & 08 06 44.41 48 41 49.2 & 08 08 33.37 42 48 36.4 & 08 10 02.06 26 03 40.5 & 08 11 36.90 28 45 03.6 + 08 11 37.23 48 31 33.8 & 08 13 18.85 50 12 39.8 & 08 14 09.22 32 37 32.0 & 08 15 40.84 39 54 37.8 & 08 18 38.88 43 17 55.6 + 08 20 50.76 40 34 56.6 & 08 21 17.15 48 45 46.3 & 08 21 25.97 51 37 15.6 & 08 21 33.61 47 02 37.3 & 08 22 14.18 30 54 37.8 + 08 23 25.27 44 58 50.5 & 08 26 16.35 52 12 09.3 & 08 30 31.93 05 20 06.8 & 08 31 36.99 30 48 26.8 & 08 32 36.34 33 31 54.7 + 08 34 44.62 34 04 14.4 & 08 38 40.58 47 34 10.7 & 08 39 23.24 06 09 58.9 & 08 42 52.40 44 34 10.6 & 08 43 52.87 37 42 28.3 + 08 47 02.79 01 30 01.4 & 08 49 25.89 56 41 32.2 & 08 50 39.95 54 37 53.4 & 08 51 14.93 01 59 53.2 & 08 52 00.44 02 29 34.5 + 08 52 32.83 38 36 56.1 & 08 53 41.19 40 52 21.8 & 09 01 02.73 42 07 46.4 & 09 02 07.20 57 07 37.9 & 09 02 21.53 48 11 36.5 + 09 02 29.23 48 39 06.2 & 09 04 36.36 51 07 28.2 & 09 05 01.56 53 39 07.5 & 09 05 13.17 42 24 51.9 & 09 06 00.09 57 47 30.1 + 09 07 45.47 38 27 39.1 & 09 08 08.80 00 35 00.5 & 09 08 12.17 51 47 00.8 & 09 08 21.02 04 50 59.5 & 09 10 17.33 37 42 52.1 + 09 12 05.16 54 31 41.3 & 09 15 28.77 44 16 32.9 & 09 17 18.68 59 03 40.4 & 09 17 57.43 02 37 34.0 & 09 18 43.09 40 17 09.7 + 09 19 21.56 50 48 55.4 & 09 22 25.14 43 07 49.4 & 09 23 08.17 56 14 55.4 & 09 23 13.54 04 34 45.0 & 09 24 54.56 49 18 04.2 + 09 28 37.98 60 25 21.0 & 09 28 56.83 07 36 19.0 & 09 31 38.29 03 15 10.1 & 09 35 18.19 02 04 15.5 & 09 35 53.82 05 03 53.2 + 09 36 28.68 01 23 29.3 & 09 41 04.01 38 53 51.0 & 09 41 24.74 03 39 16.4 & 09 41 32.54 02 04 32.6 & 09 41 44.82 57 51 23.7 + 09 42 03.05 54 05 18.9 & 09 43 58.22 02 26 30.6 & 09 44 38.79 08 51 57.0 & 09 47 40.01 51 54 56.8 & 09 51 26.48 01 46 51.8 + 09 52 28.46 06 28 10.5 & 09 52 45.57 00 00 15.5 & 09 55 56.38 06 16 42.5 & 09 58 02.82 44 06 03.7 & 10 00 54.68 53 32 06.0 + 10 03 11.56 50 57 05.2 & 10 03 50.71 52 53 52.2 & 10 05 07.08 50 19 29.6 & 10 06 33.64 44 35 00.7 & 10 09 43.56 05 29 53.9 + 10 10 27.52 41 32 39.1 & 10 11 35.45 00 57 43.1 & 10 12 54.25 61 36 34.6 & 10 13 28.77 07 56 53.8 & 10 15 41.14 59 44 45.4 + 10 16 09.93 45 31 43.2 & 10 16 51.74 @xmath700 33 47.0 & 10 21 06.04 45 23 31.9 & 10 22 35.57 45 41 05.5 & 10 23 20.89 @xmath701 02 27.5 + 10 26 31.97 06 27 33.0 & 10 27 06.26 06 09 59.2 & 10 27 25.96 52 26 37.0 & 10 30 24.95 55 16 22.8 & 10 31 43.51 52 25 35.2 + 10 33 51.88 @xmath700 24 14.3 & 10 34 18.01 08 36 26.7 & 10 38 38.82 49 47 36.9 & 10 39 36.67 07 14 27.4 & 10 41 12.53 00 45 49.3 + 10 42 07.56 50 13 22.0 & 10 43 12.84 @xmath700 13 49.1 & 10 45 34.98 @xmath700 55 29.5 & 10 45 49.00 53 47 59.8 & 10 47 26.65 01 11 47.5 + 10 47 40.47 02 07 57.3 & 10 49 32.22 05 05 31.7 & 10 51 41.17 59 13 05.2 & 10 53 10.16 58 55 32.8 & 10 53 52.87 @xmath700 58 52.8 + 10 54 54.97 06 14 53.1 & 10 54 57.04 @xmath700 45 53.0 & 10 55 00.34 52 02 00.9 & 10 55 17.27 02 05 45.1 & 10 56 54.16 05 17 13.3 + 10 57 51.58 02 27 40.9 & 11 05 14.28 01 09 08.2 & 11 07 09.51 05 47 44.8 & 11 07 15.89 05 33 06.7 & 11 07 18.88 10 04 17.7 + 11 07 45.78 60 09 13.9 & 11 12 14.45 01 20 48.5 & 11 12 17.48 61 49 26.5 & 11 17 08.86 00 43 10.6 & 11 17 16.60 57 58 11.8 + 11 17 40.84 05 28 58.7 & 11 20 48.50 03 32 47.1 & 11 24 11.66 @xmath703 37 04.7 & 11 25 06.96 @xmath700 16 47.6 & 11 27 13.99 @xmath701 11 53.3 + 11 30 44.30 01 37 24.6 & 11 32 35.86 @xmath701 28 48.7 & 11 32 50.84 @xmath700 20 01.0 & 11 33 03.03 00 15 48.9 & 11 33 25.15 55 11 24.4 + 11 36 04.92 52 25 58.9 & 11 37 03.09 01 40 06.3 & 11 37 23.46 @xmath702 15 08.2 & 11 41 11.62 @xmath701 43 06.7 & 11 43 12.04 53 50 48.3 + 11 44 33.67 60 15 38.9 & 11 45 10.39 01 10 56.3 & 11 47 34.30 04 30 47.3 & 11 48 47.83 10 54 58.4 & 11 49 18.97 02 39 26.2 + 11 51 44.39 52 44 12.6 & 11 54 05.38 56 20 40.9 & 11 54 09.28 02 38 15.0 & 11 56 51.58 05 03 35.7 & 11 58 39.92 62 54 27.9 + 11 59 23.69 01 52 24.2 & 11 59 40.97 @xmath701 27 08.5 & 12 05 17.33 00 29 56.9 & 12 07 08.02 @xmath702 44 44.1 & 12 09 49.98 54 26 31.6 + 12 11 28.87 50 52 54.2 & 12 11 54.87 60 44 26.1 & 12 15 29.56 53 35 55.9 & 12 15 41.96 05 19 32.6 & 12 15 57.60 @xmath700 07 19.8 + 12 17 01.37 10 19 52.9 & 12 18 58.76 01 52 37.9 & 12 20 27.98 09 28 27.3 & 12 20 49.15 58 59 21.6 & 12 21 30.39 @xmath702 41 33.0 + 12 31 24.11 01 12 07.3 & 12 36 23.80 06 02 08.2 & 12 36 33.12 10 09 28.7 & 12 37 53.86 00 16 37.8 & 12 38 29.96 @xmath700 13 27.9 + 12 40 00.93 03 40 51.6 & 12 41 16.49 51 41 30.0 & 12 41 39.73 49 34 05.5 & 12 41 57.55 63 32 41.6 & 12 42 16.38 00 47 43.7 + 12 44 47.32 59 41 07.4 & 12 45 38.35 55 11 32.7 & 12 47 10.29 55 55 57.3 & 12 47 58.52 62 50 49.3 & 12 51 51.03 49 18 55.0 + 12 53 11.26 01 20 27.3 & 12 54 02.16 @xmath700 49 31.1 & 12 55 08.10 62 20 50.5 & 12 55 28.30 @xmath700 54 42.0 & 12 55 54.74 57 54 25.4 + 12 57 03.12 00 24 36.0 & 12 57 29.79 @xmath701 32 39.6 & 12 57 37.07 @xmath700 32 20.2 & 12 58 24.70 02 08 46.8 & 12 59 45.18 03 17 26.2 + 13 02 10.16 05 08 15.2 & 13 05 21.33 49 51 42.3 & 13 08 28.73 50 26 23.2 & 13 08 42.88 02 43 26.7 & 13 09 07.99 52 24 37.3 + 13 10 28.51 00 44 08.9 & 13 10 40.56 @xmath703 34 12.0 & 13 10 40.74 02 01 27.1 & 13 15 38.72 @xmath701 58 46.2 & 13 16 00.79 @xmath702 18 19.6 + 13 16 14.50 02 39 38.8 & 13 17 26.13 @xmath702 31 50.5 & 13 18 27.00 62 00 36.3 & 13 25 09.70 00 49 33.9 & 13 26 07.16 47 54 41.5 + 13 26 31.45 47 37 55.9 & 13 26 55.72 02 07 27.4 & 13 27 46.16 48 42 03.0 & 13 29 09.25 48 01 09.7 & 13 32 21.85 53 28 17.4 + 13 32 59.17 49 09 46.8 & 13 34 11.70 55 01 25.0 & 13 34 37.49 56 31 47.9 & 13 38 02.81 42 39 57.0 & 13 39 48.45 47 41 17.1 + 13 40 42.88 02 03 07.5 & 13 40 48.37 43 33 59.9 & 13 41 34.85 53 44 43.7 & 13 45 45.36 53 32 52.3 & 13 46 17.55 62 20 45.5 + 13 47 39.84 62 21 49.6 & 13 50 54.59 05 22 06.5 & 13 52 23.24 48 46 12.2 & 13 53 05.54 04 43 38.7 & 13 54 09.95 @xmath701 41 50.7 + 13 57 03.83 02 30 07.1 & 13 58 23.52 60 25 07.3 & 13 59 53.80 59 11 03.0 & 13 59 59.08 @xmath701 44 54.2 & 14 01 30.69 41 55 15.3 + 14 02 14.80 58 17 46.8 & 14 05 18.48 04 34 06.9 & 14 06 56.49 46 17 12.5 & 14 08 32.48 47 38 37.4 & 14 08 32.65 00 31 38.5 + 14 10 30.99 61 41 37.0 & 14 10 54.06 58 46 55.4 & 14 11 23.51 00 42 53.0 & 14 12 31.18 54 55 11.5 & 14 15 32.19 40 49 51.6 + 14 17 08.16 46 07 05.4 & 14 17 40.03 60 53 23.8 & 14 19 32.59 00 31 20.3 & 14 20 33.26 @xmath700 32 33.3 & 14 21 27.77 58 47 57.6 + 14 22 35.89 @xmath701 52 11.2 & 14 24 14.09 42 14 00.1 & 14 26 06.19 40 24 32.0 & 14 27 45.71 55 04 10.3 & 14 27 46.79 00 28 44.7 + 14 28 29.93 44 39 49.7 & 14 30 17.34 52 17 35.1 & 14 32 44.44 @xmath700 59 15.1 & 14 33 39.38 50 52 26.9 & 14 34 10.77 @xmath701 23 41.7 + 14 36 27.37 52 14 00.0 & 14 37 56.93 01 56 38.9 & 14 38 06.25 @xmath700 35 34.8 & 14 38 20.93 @xmath702 39 53.1 & 14 38 34.54 39 42 60.0 + 14 39 39.95 44 28 51.0 & 14 40 01.64 61 01 33.9 & 14 43 34.57 @xmath702 19 26.5 & 14 44 14.21 43 36 55.5 & 14 46 36.90 00 46 56.5 + 14 47 07.41 52 03 40.1 & 14 48 14.94 57 41 45.7 & 14 50 49.93 00 01 44.4 & 14 51 34.61 01 59 37.0 & 14 52 40.92 43 08 14.3 + 14 52 47.37 47 35 29.2 & 15 00 07.27 56 36 00.8 & 15 00 31.81 48 36 46.8 & 15 01 21.97 01 44 01.2 & 15 04 05.11 46 28 51.3 + 15 04 31.30 47 41 51.3 & 15 05 45.05 41 17 26.7 & 15 08 24.74 56 04 23.3 & 15 08 35.94 60 32 58.6 & 15 09 40.59 60 38 21.3 + 15 11 09.22 56 50 51.8 & 15 12 15.74 02 03 17.0 & 15 13 25.62 42 00 16.8 & 15 13 34.31 46 12 57.7 & 15 13 56.16 04 20 55.8 + 15 14 15.96 57 49 04.0 & 15 15 03.23 61 35 20.1 & 15 19 36.72 53 42 55.5 & 15 20 21.04 02 53 12.1 & 15 20 25.64 59 39 55.4 + 15 23 46.79 @xmath702 22 39.5 & 15 25 05.36 53 06 22.9 & 15 26 03.80 58 25 25.1 & 15 28 38.40 48 47 40.7 & 15 29 24.65 @xmath701 53 43.5 + 15 31 07.41 58 44 09.9 & 15 31 28.69 03 38 02.3 & 15 43 16.52 41 36 09.5 & 15 44 30.48 41 20 14.0 & 15 44 39.71 44 40 51.0 + 15 52 21.45 00 12 56.9 & 15 54 03.09 32 33 32.6 & 15 57 52.76 02 53 27.9 & 16 03 26.34 49 30 44.3 & 16 05 56.55 26 59 44.2 + 16 08 00.02 38 15 30.7 & 16 08 46.76 37 48 50.7 & 16 13 42.98 39 07 32.8 & 16 13 51.34 37 42 58.7 & 16 22 29.93 35 31 25.4 + 16 23 36.45 34 19 46.4 & 16 24 21.99 39 24 40.9 & 16 25 13.80 40 58 51.0 & 16 28 51.27 45 52 18.5 & 16 29 17.79 44 34 52.4 + 16 29 57.80 42 30 51.4 & 16 30 46.21 36 13 06.0 & 16 35 22.86 39 44 37.8 & 16 35 24.18 31 10 00.4 & 16 36 09.24 26 23 09.1 + 16 36 24.31 47 15 35.8 & 16 36 40.56 46 47 07.2 & 16 36 54.41 32 20 06.5 & 16 37 02.21 41 30 22.2 & 16 38 56.54 43 35 12.6 + 16 39 55.99 47 05 23.6 & 16 40 54.16 31 43 29.9 & 16 44 52.58 37 30 09.3 & 16 45 44.69 37 55 26.1 & 16 46 34.71 35 03 17.6 + 16 50 27.67 36 22 56.5 & 16 58 19.55 62 38 23.2 & 16 58 20.35 37 33 15.6 & 16 59 12.68 40 03 59.1 & 16 59 43.08 37 54 22.7 + 17 01 23.98 38 51 37.1 & 17 03 35.00 39 17 35.6 & 17 04 04.50 38 54 30.7 & 17 04 49.89 31 17 28.2 & 17 05 18.52 35 33 52.3 + 17 06 48.06 32 14 22.9 & 17 06 51.42 38 16 45.3 & 17 08 01.25 33 46 46.3 & 17 13 28.81 30 59 07.8 & 17 14 30.10 61 57 46.6 + 17 15 54.63 28 44 49.9 & 17 20 51.15 62 09 44.3 & 17 21 58.61 55 47 07.5 & 21 07 54.98 @xmath706 25 19.1 & 21 08 18.46 @xmath706 17 40.5 + 21 27 15.34 @xmath706 20 41.7 & 21 30 04.76 @xmath701 02 44.4 & 21 35 13.10 @xmath700 52 43.9 & 21 41 11.89 @xmath706 39 30.3 & 21 44 32.75 @xmath707 54 42.8 + 21 49 37.38 @xmath706 53 12.9 & 21 52 13.51 @xmath707 42 24.9 & 21 59 34.46 @xmath709 10 22.2 & 22 07 52.53 @xmath700 17 35.8 & 22 14 09.97 00 52 27.0 + 22 14 26.04 @xmath708 16 37.7 & 22 27 20.56 @xmath700 01 59.3 & 22 27 29.06 00 05 22.0 & 22 29 12.25 @xmath709 42 18.9 & 22 55 10.38 @xmath709 07 55.1 + 23 05 45.67 @xmath700 36 08.6 & 23 12 12.16 @xmath709 19 28.7 & 23 20 20.78 @xmath700 20 10.5 & 23 21 33.77 @xmath701 06 46.0 & 23 35 34.68 @xmath709 27 39.2 + 23 41 42.32 00 33 12.6 & 23 45 40.45 @xmath709 36 10.3 & 23 47 24.71 00 52 47.0 & 23 50 06.26 @xmath700 09 33.4 & 23 50 26.40 @xmath710 19 58.1 + 23 51 56.13 @xmath701 09 13.4 & 23 53 21.04 @xmath708 59 30.6 + lrlrlr [ fr2sampleprop ] @xmath72 & 1.90 & @xmath73 & 171.3 & @xmath74 & 1.476 + @xmath75 & 1.92 & @xmath76 & 171.7 & @xmath77 & 1.411 + @xmath78 & 1.88 & @xmath79 & 170.9 & @xmath80 & 1.540 + @xmath81 & 1.48 & @xmath82 & 173.2 & @xmath83 & 0.859 + @xmath84 & 1.52 & @xmath85 & 173.6 & @xmath86 & 0.940 + @xmath87 & 1.44 & @xmath88 & 172.7 & @xmath89 & 0.813 + lrcrrrcrcrrrcrcrrr [ lineprop ] ly@xmath30 & 1216.5 & 15.4 & 74.50 & 100.0 & 406.7 & & 1215.7 & 13.2 & 81.20 & 100.0 & 347.0 & & 1218.6 & 29.7 & 75.22 & 100.0 & 428.9 + c iv & 1546.8 & 21.0 & 35.81 & 35.55 & 144.6 & & 1545.9 & 19.7 & 22.44 & 21.10 & 73.21 & & 1545.5 & 24.1 & 27.40 & 26.94 & 115.6 + he ii & 1637.9 & 11.2 & 2.31 & 2.134 & 8.679 & & 1637.3 & 21.3 & 2.54 & 2.239 & 7.769 & & 1636.5 & 21.6 & 2.47 & 2.259 & 9.688 + o iii@xmath60 $ ] & 1661.4 & 13.4 & 1.65 & 1.497 & 6.089 & & 1664.4 & 19.6 & 1.22 & 1.056 & 3.663 & & 1664.2 & 17.0 & 0.75 & 0.671 & 2.880 + al iii & 1858.2 & 35.8 & 1.50 & 1.183 & 4.810 & & 1858.1 & 15.1 & 1.85 & 1.415 & 4.909 & & 1857.7 & 26.9 & 2.07 & 1.612 & 6.913 + c iii@xmath60 $ ] & 1905.6 & 29.8 & 18.45 & 14.09 & 57.32 & & 1905.9 & 33.5 & 19.54 & 14.52 & 50.39 & & 1905.0 & 37.0 & 20.99 & 15.83 & 67.89 + mg ii ( b ) & 2797.8 & 106.0 & 29.04 & 13.69 & 55.69 & & 2799.3 & 104.7 & 24.95 & 12.04 & 41.79 & & 2797.9 & 126.6 & 22.51 & 10.42 & 44.70 + mg ii ( n ) & 2797.8 & 23.2 & 8.81 & 4.154 & 16.90 & & 2799.8 & 24.2 & 7.75 & 3.740 & 12.98 & & 2800.4 & 25.9 & 12.52 & 5.790 & 24.84 + @xmath61ne v@xmath60 $ ] & 3344.8 & 9.0 & 0.60 & 0.226 & 0.920 & & 3345.1 & 9.9 & 0.42 & 0.166 & 0.576 & & 3344.1 & 8.4 & 0.33 & 0.121 & 0.523 + @xmath61ne v@xmath60 $ ] & 3426.3 & 6.8 & 1.87 & 0.684 & 2.780 & & 3424.4 & 7.9 & 1.02 & 0.393 & 1.362 & & 3423.3 & 13.1 & 1.33 & 0.477 & 2.045 + @xmath61o ii@xmath60 $ ] & 3728.6 & 8.5 & 2.27 & 0.746 & 3.035 & & 3728.7 & 9.4 & 2.11 & 0.738 & 2.561 & & 3729.2 & 8.2 & 1.29 & 0.415 & 1.779 + @xmath61ne iii@xmath60 $ ] & 3869.2 & 7.6 & 1.49 & 0.468 & 1.902 & & 3868.7 & 9.0 & 1.88 & 0.631 & 2.189 & & 3868.3 & 10.1 & 1.49 & 0.457 & 1.962 + h@xmath90 & 4101.0 & 29.3 & 3.73 & 1.088 & 4.425 & & 4098.0 & 62.0 & 3.50 & 1.101 & 3.821 & & 4102.5 & 46.4 & 5.58 & 1.590 & 6.818 + h@xmath91 & 4340.6 & 17.2 & 9.02 & 2.450 & 9.964 & & 4339.8 & 33.7 & 8.12 & 2.395 & 8.311 & & 4340.9 & 32.8 & 12.46 & 3.304 & 14.17 + @xmath61o iii@xmath60 $ ] & 4364.4 & 8.0 & 5.44 & 1.468 & 5.968 & & 4364.3 & 26.7 & 6.98 & 2.046 & 7.100 & & 4363.6 & 14.2 & 6.13 & 1.615 & 6.926 + h@xmath92 ( b ) & 4862.1 & 66.1 & 20.20 & 4.758 & 19.35 & & 4862.4 & 43.3 & 23.52 & 6.107 & 21.19 & & 4862.5 & 40.0 & 32.12 & 7.375 & 31.63 + h@xmath92 ( n ) & 4862.1 & 7.8 & 2.30 & 0.542 & 2.203 & & 4862.4 & 6.9 & 1.62 & 0.421 & 1.459 & & 4862.5 & 7.1 & 1.67 & 0.384 & 1.645 + @xmath61o iii@xmath60 $ ] & 4959.7 & 7.5 & 6.62 & 1.521 & 6.185 & & 4959.6 & 10.2 & 3.76 & 0.955 & 3.313 & & 4959.6 & 9.7 & 3.53 & 0.791 & 3.390 + @xmath61o iii@xmath60 $ ] & 5007.6 & 7.2 & 21.17 & 4.805 & 19.54 & & 5007.5 & 9.2 & 13.61 & 3.419 & 11.86 & & 5007.7 & 8.9 & 12.37 & 2.736 & 11.74 + @xmath61o i@xmath60 $ ] & 6301.5 & 17.3 & 1.95 & 0.332 & 1.349 & & 6301.0 & 14.0 & 1.96 & 0.380 & 1.320 & & 6302.9 & 9.7 & 1.15 & 0.190 & 0.815 + h@xmath93 & 6561.3 & 58.4 & 95.02 & 15.36 & 62.47 & & 6564.9 & 22.8 & 159.52 & 29.56 & 102.6 & & 6564.6 & 20.6 & 146.55 & 22.99 & 98.62 + @xmath61n ii@xmath60 $ ] & 6584.3 & 11.1 & 8.87 & 1.428 & 5.806 & & 6586.1 & 12.2 & 12.71 & 2.347 & 8.144 & & 6585.9 & 8.6 & 7.66 & 1.197 & 5.133 + @xmath61s ii@xmath60 $ ] & 6718.7 & 8.2 & 1.64 & 0.257 & 1.047 & & 6719.5 & 8.8 & 4.59 & 0.829 & 2.875 & & 6718.8 & 8.2 & 2.55 & 0.388 & 1.666 + @xmath61s ii@xmath60 $ ] & 6732.0 & 8.1 & 0.93 & 0.146 & 0.592 & & 6734.2 & 6.7 & 4.45 & 0.802 & 2.781 & & 6733.3 & 8.8 & 2.01 & 0.305 & 1.310 +	we have combined a sample of 44984 quasars , selected from the sloan digital sky survey ( sdss ) data release 3 , with the first radio survey . using a novel technique where the optical quasar position is matched to the complete radio _ environment _ within 450 , we are able to characterize the radio morphological make - up of what is essentially an optically selected quasar sample , regardless of whether the quasar ( nucleus ) itself has been detected in the radio . about 10% of the quasar population have radio cores brighter than 0.75 mjy at 1.4 ghz , and 1.7% have double lobed fr2-like radio morphologies . about 75% of the fr2 sources have a radio core ( @xmath0mjy ) . a significant fraction ( @xmath1% ) of the fr2 quasars are bent by more than 10 degrees , indicating either interactions of the radio plasma with the icm or igm . we found no evidence for correlations with redshift among our fr2 quasars : radio lobe flux densities and radio source diameters of the quasars have similar distributions at low ( mean 0.77 ) and high ( mean 2.09 ) redshifts . using a smaller high reliability fr2 sample of 422 quasars and two comparison samples of radio - quiet and non - fr2 radio - loud quasars , matched in their redshift distributions , we constructed composite optical spectra from the sdss spectroscopic data . based on these spectra we can conclude that the fr2 quasars have stronger high - ionization emission lines compared to both the radio quiet and non - fr2 radio loud sources . this is consistent with the notion that the emission lines are brightened by ongoing shock ionization of ambient gas in the quasar host as the radio source expands .
let us consider an ordinary differential equation ( ode ) @xmath0 of order @xmath1 with respect to the unknown @xmath2 , where @xmath3 is a polynomial of @xmath4 variables , @xmath5 . the classical maillet theorem @xcite asserts that any formal power series solution @xmath6 $ ] of ( [ ade ] ) is a power series of gevrey order @xmath7 for some @xmath8 . this means that the power series @xmath9 converges in a neighbourhood of zero , where @xmath10 is the euler gamma - function . in other words , @xmath11 for some @xmath12 . first exact estimates for the gevrey order of a power series formally satisfying an ode were obtained by j .- p.ramis @xcite in the linear case , @xmath13 he has proved that such a power series is of the _ exact _ gevrey order @xmath14 , where @xmath15 are all of the positive slopes of the newton polygon @xmath16 of the operator @xmath17 . this polygon is defined as the boundary curve of a convex hull of a union @xmath18 , @xmath19 ( see fig . the exactness of the gevrey order means that there is no @xmath20 such that the power series is of the gevrey order @xmath21 . with two positive slopes @xmath22 @xmath23.,height=151 ] the result of ramis has been further generalized by b.malgrange and y.sibuya for a non - linear ode of the general form ( [ ade ] ) . * theorem 1 * ( malgrange @xcite ) . _ let @xmath24 $ ] satisfy the equation @xmath25 , that is @xmath26 , where @xmath27 , and @xmath28 . then @xmath29 is a power series of gevrey order @xmath7 , where @xmath30 is the least of all the positive slopes of the newton polygon @xmath31 of a linear operator @xmath32 @xmath33or @xmath34 , if @xmath31 has no positive slopes@xmath35 . _ the refinement of theorem 1 belongs to y.sibuya ( * ? ? ? this claims that @xmath29 is a power series of the _ exact _ gevrey order @xmath14 , where @xmath15 are all of the positive slopes of the newton polygon @xmath31 . in the paper we study _ generalized _ power series solutions of ( [ ade ] ) of the form @xmath36 with the power exponents satisfying conditions @xmath37 ( the latter , in particular , implies that a set of exponents having a fixed real part is finite ) . note that substituting the series ( [ series ] ) into the equation ( [ ade ] ) makes sense , as only a finite number of terms in @xmath29 contribute to any term of the form @xmath38 in the expansion of @xmath39 in powers of @xmath40 . indeed , @xmath41 and an equation @xmath42 has a finite number of solutions @xmath43 , since @xmath44 . furthermore , for any integer @xmath45 an inequality @xmath46 has also a finite number of solutions , so that powers of @xmath40 in the expansion of @xmath47 can be ordered by the increasing of real parts . thus , one may correctly define the notion of a formal solution of ( [ ade ] ) in the form of a generalized power series . in particular , the painlev iii , v , vi equations are known to have such formal solutions ( see @xcite , @xcite , @xcite , @xcite ) . for the generalized power series ( [ series ] ) one may naturally define the _ valuation _ @xmath48 and this is also well defined for any polynomial in @xmath49 . the main result of the paper is an analogue of the maillet theorem ( more precisely , of the malgrange theorem ) for generalized power series . * theorem 2 . * _ let the generalized power series @xmath50 formally satisfy the equation @xmath25 , @xmath28 , and for each @xmath51 one have @xmath52 where not all @xmath53 equal zero . let @xmath30 be the least of all the positive slopes of the newton polygon @xmath31 @xmath33or @xmath34 , if @xmath31 has no positive slopes@xmath35 . then for any sector @xmath54 of sufficiently small radius with the vertex at the origin and of the opening less than @xmath55 , the series @xmath56 converges uniformly in @xmath54 . _ the newton polygon of @xmath57 in the case of the generalized power series @xmath29 is defined similarly to the classical case , as the boundary curve of a convex hull of a union @xmath58 note that @xmath34 in theorem 2 if and only if @xmath59 . in this case @xmath29 does converge in @xmath54 , which has been already proved in @xcite by the majorant method . here we consider the case @xmath60 using other known methods rather than the majorant one ( of course , the convergence could also be proved by these methods ) . the first step ( sections 2 , 3 ) consists of representing the generalized power series solution ( [ series ] ) of ( [ ade ] ) by a multivariate taylor series and follows from the grid - basedness of this formal solution . the latter means that there are @xmath61 such that any term @xmath38 of @xmath29 is of the form @xmath62 , for some @xmath63 . an earlier result of this kind is due to grigoriev , singer @xcite studying generalized power series solutions with real exponents . according to later results @xcite , @xcite , more general ( real ) transseries solutions of a polynomial ode are actually also grid - based ( see @xcite as an introduction to transseries ) . the second step ( section 4 ) consists of applying the implicit mapping theorem for banach spaces of the obtained multivariate taylor series . as we mentioned above , we prove theorem 2 in the case @xmath60 . this implies @xmath64 in the expansion ( [ condition ] ) of @xmath65 . let @xmath66 be such that @xmath67 and @xmath68 for all @xmath69 . then the minimal positive slope @xmath30 of the newton polygon @xmath31 is @xmath70 first we will reduce the equation ( [ ade ] ) to a special form . this is provided by a transformation @xmath71 with @xmath72 that will be chosen later . we use ideas of @xcite adapted to the case of generalized power series . the formal solution ( [ series ] ) can be represented in the form @xmath73 taking into consideration the equality @xmath74 , we have the relations @xmath75 therefore , denoting @xmath76 and applying the taylor formula to the relation @xmath26 , we have @xmath77 according to the assumptions of theorem 2 and definition of the integer @xmath78 , for each @xmath51 the formal series @xmath79 is of the form @xmath80 where @xmath68 for all @xmath69 and @xmath81 . define a polynomial @xmath82 of degree @xmath78 and choose an integer @xmath72 such that the following three conditions i ) , ii ) , iii ) hold . * @xmath83 with @xmath84 ; * @xmath85 ; * @xmath86 . note that for each @xmath51 , the real part of @xmath87 is greater than @xmath88 . therefore ( see ( [ condition ] ) for @xmath79 ) , @xmath89 moreover , for @xmath69 we have @xmath90 indeed , if @xmath91 , then the above inequality follows from the definition ( [ slope ] ) of the slope @xmath30 , otherwise one has @xmath92 , and ( [ tildelambda ] ) follows from the condition iii ) above . thus , ( [ decomposition ] ) and ( [ tildelambda ] ) imply a decomposition @xmath93 where the polynomial @xmath17 is defined by the formula ( [ euler ] ) , and exponents @xmath94 in the monomials @xmath95 of the operator @xmath96 satisfy the inequalities @xmath97 from the relation ( [ taylor ] ) , condition ii ) and inequality @xmath88 it follows that @xmath98 finally , dividing the relation ( [ taylor ] ) by @xmath99 and using ( [ ll ] ) , ( [ valf ] ) , and the condition iii ) , we obtain the equality of the form @xmath100 where the linear differential operators @xmath101 , @xmath96 are described above , and @xmath102 is a finite linear combination of monomials of the form @xmath103 thus , the transformation @xmath104 reduces the equation ( [ ade ] ) to an equation @xmath105 with a formal solution @xmath106 . * remark 1 . * the condition i ) is not used for the above reduction , but we add it from the beginning as this will be used in the sequel . in particular , i ) implies that the coefficients @xmath107 , @xmath108 , of the formal solution @xmath109 of ( [ reduced ] ) are uniquely determined by the coefficients @xmath110 of the partial sum @xmath111 of @xmath29 ( whereas some of @xmath110 themselves may be free parameters ) . let us define an additive semi - group @xmath112 generated by a ( finite ) set consisting of the number @xmath113 and all the power exponents @xmath94 , @xmath114 of the variable @xmath40 containing in the monomials @xmath95 , @xmath115 of @xmath96 , @xmath102 respectively . let @xmath116 be generators of this semi - group , that is , @xmath117 as a consequence of the relation ( [ psirelation ] ) and condition that the real parts of the power exponents @xmath94 , @xmath114 are positive , we have the following auxiliary lemma ( details of the proof see in ( * ? ? ? * lemma 2 ) ) . * lemma 1 . * _ all the numbers @xmath118 , @xmath108 , belong to the additive semi - group @xmath112 . _ we may assume that the generators @xmath116 of @xmath112 are linearly independent over @xmath119 . this is provided by the following lemma ( * ? ? ? * lemma 3 ) . * lemma 2 . * _ there are complex numbers @xmath120 linearly independent over @xmath119 , such that all @xmath121 and an additive semi - group @xmath122 generated by them contains the above semi - group @xmath112 generated by @xmath116 . _ in what follows , for the simplicity of exposition we assume that @xmath112 is generated by two numbers : @xmath123 in the case of an arbitrary number @xmath124 of generators all constructions are analogous , only multivariate taylor series in @xmath124 rather than in two variables are involved . we should estimate the growth of the coefficients @xmath107 of the generalized power series @xmath125 which satisfies the equality ( [ psirelation ] ) . according to lemma 1 , all the exponents @xmath118 belong to the semi - group @xmath112 : @xmath126 for some set @xmath127 such that the map @xmath128 is a bijection from @xmath129 to @xmath127 . hence , @xmath130 ( in the last series one puts @xmath131 , if @xmath132 ) . now we define a natural linear map @xmath133\rightarrow{\mathbb c}[[z_1,z_2]]_$ ] from the @xmath134-algebra of generalized power series with exponents in @xmath112 to the @xmath134-algebra of taylor series in two variables without a constant term , @xmath135 as follows from the linear independence of the generators @xmath136 over @xmath119 , @xmath137,\ ] ] hence @xmath138 is an isomorphism . the differentiation @xmath139\rightarrow{\mathbb c}[[z^g]]$ ] naturally induces a linear bijective map @xmath140 of @xmath141_$ ] to itself , @xmath142 which clearly satisfies @xmath143 , so that the following commutative diagramme holds : @xmath144 & \stackrel{\delta}{\longrightarrow } & { \mathbb c}[[z^g ] ] \\ \downarrow\lefteqn{\sigma } & & \downarrow\lefteqn{\sigma } \\ { \mathbb c}[[z_1,z_2 ] ] _ * & \stackrel{\delta}{\longrightarrow } & { \mathbb c}[[z_1,z_2 ] ] _ * \end{array}\ ] ] thus we have the representation @xmath145 of the formal solution @xmath146 of ( [ reduced ] ) by a multivariate taylor series , where @xmath147 for every @xmath148 . now we apply the map @xmath138 to the both sides of the equality ( [ psirelation ] ) , @xmath149 and obtain a relation for @xmath150 : @xmath151 where @xmath152 , and * @xmath153 is a finite linear combination of monomials @xmath154 which satisfy @xmath155 furthermore @xmath156 ; * @xmath157 is a polynomial , and @xmath158 ; * @xmath159 . in this section we conclude the proof of theorem 2 establishing the corresponding required properties for the multivariate taylor series @xmath150 , which represents the generalized power series @xmath146 and satisfies the relation ( [ psi0relation ] ) . we use the dilatation method based on the implicit mapping theorem for banach spaces . this was originally used by malgrange for proving theorem 1 , as well as by c.zhang @xcite in a further generalization of this theorem for @xmath160-difference - differential equations . let us define the following banach spaces @xmath161 of ( formal ) taylor series in two variables without a constant term : @xmath162 with the norm @xmath163 ( the completeness of each @xmath161 is checked in a way similar to that how one checks the completeness of the space @xmath164 ; see , for example , ( * ? ? ? * ch . 6 , 4 ) . ) one clearly has @xmath165 and @xmath166 therefore , @xmath167 maps @xmath168 to @xmath169 , whereas @xmath170 for @xmath69 may map @xmath168 outside @xmath169 . however , linear operators @xmath154 with suitable @xmath171 , @xmath164 possess the following property . * proof . * let @xmath174 . then , by definition , @xmath175 therefore @xmath176 where @xmath177 is an absolutely convergent series . thus we have @xmath178 where we write @xmath179 for @xmath148 , @xmath180 , if @xmath181 , @xmath182 and @xmath183 . to check the inclusion @xmath184 , we should prove the absolute convergence of the series @xmath185 since for @xmath186 , @xmath187 , and a fixed @xmath188 , one has ( see ( * ? ? ? i , 1.4 ) ) @xmath189 a general term of the series ( [ h0 ] ) is equivalent to @xmath190 . this implies the convergence of this series , as @xmath191 under the assumptions of the lemma . we also have @xmath192 whence the continuity of @xmath154 follows . _ let @xmath207 , @xmath208 , @xmath209 be banach spaces , @xmath210 an open subset of the direct product @xmath211 and @xmath212 a continuously differentiable mapping . consider a point @xmath213 such that @xmath214 and @xmath215 is a bijective linear mapping from @xmath208 to @xmath209 . _ we will apply this theorem to the banach spaces @xmath134 , @xmath168 , @xmath169 , and to the mapping @xmath223 defined by @xmath224 with @xmath17 , @xmath225 , and @xmath226 coming from ( [ psi0relation ] ) . this mapping is continuously differentiable by lemmas 3 , 4 , moreover @xmath227 and @xmath228 is a bijective linear mapping from @xmath168 to @xmath169 . therefore , there are @xmath229 and @xmath230 such that @xmath231 making the change of variables @xmath232 , which induces an automorphism @xmath233 of @xmath141_$ ] commuting with @xmath140 , one can easily see that the above relation implies that the power series @xmath234 satisfies the same equality ( [ psi0relation ] ) as @xmath235 does . hence , these two series coincide ( the coefficients of a series satisfying ( [ psi0relation ] ) are determined uniquely by this equality ) and @xmath236 has a non - zero radius of convergence . this implies ( substitute @xmath237 , @xmath238 remembering that @xmath239 ) the convergence of the series @xmath240 for any @xmath40 from a sector @xmath54 of sufficiently small radius with the vertex at the origin and of the opening less than @xmath55 , whence theorem 2 follows . a.parusnikova , asymptotic expansions of solutions to the fifth painlev equation in neighbourhoods of singular and nonsingular points of the equation , in : formal and analytic solutions of differential and difference equations , _ banach center publ . _ 97 * ( 2012 ) , 113124 .	there is proposed the maillet malgrange type theorem for a generalized power series ( having complex power exponents ) formally satisfying an algebraic ordinary differential equation . the theorem describes the growth of the series coefficients . * the maillet malgrange type theorem for generalized power series * * r.r.gontsov , i.v.goryuchkina *
neutrinos are very elusive particles that are difficult to detect . even so , physicists have over the last decades successfully studied neutrinos from a wide variety of sources , either natural , such as the sun and cosmic objects , or manmade , such as nuclear power plants or accelerated beams . spectacular results have been obtained in the last few years for the flavour mixing of neutrinos obtained from atmospheric , solar , reactor and accelerator sources and interpreted in terms of the survival probabilities for the beautiful quantum phenomenon of neutrino oscillations @xcite . the weak interaction eigenstates @xmath12 @xmath13 are written in terms of mass eigenstates @xmath14 @xmath15 as @xmath16 , where @xmath17 are the mixing angles among the three neutrino families and @xmath8 is the @xmath18 violating phase . neutrino mass differences and the mixings for the atmospheric @xmath19 and solar @xmath20 sectors have thus been determined . the third connecting mixing @xmath21 is bounded as @xmath22 from the chooz reactor experiment @xcite . the third angle , @xmath23 , as well as the cp - violating phases @xmath8 , remain thus undetermined . besides the approved experiments double chooz @xcite , t2k @xcite and nova @xcite , a number of experimental facilities to significantly improve on present sensitivity have been discussed in the literature : neutrino factories ( neutrino beams from boosted - muon decays ) @xcite , superbeams ( very intense conventional neutrino beams ) @xcite improved reactor experiments @xcite and more recently @xmath24-beams @xcite . the original standard scenario for beta beams with lower @xmath25 and short baseline @xmath4 km from cern to frejus with @xmath26 and @xmath27 ions could be improved using an electron capture facility for monochromatic neutrino beams @xcite . new proposals also include the high @xmath28 value @xmath29 and @xmath30 isotopes in a @xmath31 facility @xcite . in this paper we discuss the physics reach that a high energy facility for both beta beams @xcite and ec beams may provide with the expected sps upgrade at cern . in section 2 we discuss the virtues of the suppressed oscillation channel @xmath32 in order to have access to the parameters @xmath23 and @xmath8 . the interest of energy dependence , as obtainable in the ec facility , is emphasized . in section 3 we compare the beta beam capabilities at different energies and baselines using two ions , one for neutrinos , the other for antineutrinos . in section 4 we present new results on the comparison between ( low energies , short baseline ) and ( high energies , long baseline ) configurations for an ec facility with a single ion . section 5 gives our conclusions and outlook . the observation of @xmath18 violation needs an experiment in which the emergence of another neutrino flavour is detected rather than the deficiency of the original flavour of the neutrinos . at the same time , the interference needed to generate cp - violating observables can be enhanced if both the atmospheric and solar components have a similar magnitude . this happens in the suppressed @xmath33 transition . the appearance probability @xmath34 as a function of the distance between source and detector @xmath35 is given by @xcite @xmath36 where @xmath37 . the three terms of eq . ( [ prob ] ) correspond , respectively , to contributions from the atmospheric and solar sectors and their interference . as seen , the @xmath18 violating contribution has to include all mixings and neutrino mass differences to become observable . the four measured parameters @xmath38 and @xmath39 have been fixed throughout this paper to their mean values @xcite . neutrino oscillation phenomena are energy dependent ( see fig.[proba ] ) for a fixed distance between source and detector , and the observation of this energy dependence would disentangle the two important parameters : whereas @xmath21 gives the strength of the appearance probability , the @xmath18 phase acts as a phase - shift in the interference pattern . these properties suggest the consideration of a facility able to study the detailed energy dependence by means of fine tuning of a boosted monochromatic neutrino beam . as shown below , in an electron capture facility the neutrino energy is dictated by the chosen boost of the ion source and the neutrino beam luminosity is concentrated at a single known energy which may be chosen at will for the values in which the sensitivity for the @xmath40 parameters is higher . this is in contrast to beams with a continuous spectrum , where the intensity is shared between sensitive and non sensitive regions . furthermore , the definite energy would help in the control of both the systematics and the detector background . in the beams with a continuous spectrum , the neutrino energy has to be reconstructed in the detector . in water - cerenkov detectors , this reconstruction is made from supposed quasielastic events by measuring both the energy and direction of the charged lepton . this procedure suffers from non - quasielastic background , from kinematic deviations due to the nuclear fermi momentum and from dynamical suppression due to exclusion effects @xcite . from general arguments of cpt invariance and absence of absorptive parts the cp - odd probability is odd in time and then odd in the baseline l ( formally ) . vacuum oscillations are only a function of @xmath41 so that , at fixed @xmath42 , the cp - odd probability is odd in the energy ( formally ) . this proves that the study of neutrino oscillations in terms of neutrino energy will be able to separate out the cp phase @xmath8 from the mixing parameters . a control of this energy may be obtained from the choice of the boost in the ec facility with a single ion . in order for this concept to become operational it is necessary to combine it with the recent discovery of nuclei far from the stability line , having super allowed spin - isospin transitions to a giant gamow - teller resonance kinematically accessible @xcite . thus the rare - earth nuclei above @xmath43 have a small enough half - life for electron capture processes . this is in contrast with the proposal of ec beams with fully stripped long - lived ions @xcite . we discuss the option of short - lived ions @xcite . a first question to be answered is : is the sensitivity to cp violation and @xmath23 changing with energy at fixed baseline ? fixing the baseline to cern - frejus ( @xmath4 km ) @xcite , one notices that , for @xmath44 , the sensitivity to both @xmath23 and @xmath8 changes rather slowly because the flux at low energies in the continuous spectrum does not reduce significantly . then it is not advantageous to increase the neutrino energy unless the baseline is correspondingly scaled to remain close to the atmospheric oscillation maximum as suggested by the @xmath41 dependence . with the present sps of cern the maximum energy reachable for the @xmath26 ion corresponds to @xmath45 . fixing this value of @xmath46 for both @xmath26 and @xmath27 we may ask : is the sensitivity to @xmath23 and @xmath8 changing with the baseline ? particularly for the cp phase @xmath8 , @xmath47 km is clearly favoured @xcite . there is neither an existing nor an envisaged laboratory at this particular distance from cern . equipped with these previous results , it is of interest to make a comparison between the physics reach for two different beta beam setups . with the same @xmath46 for both neutrino and antineutrino sources , setup i corresponds to @xmath3 , @xmath4 km ( frejus ) , whereas setup ii is for @xmath5 , @xmath6 km ( canfranc ) . setup ii needs the upgrade of the sps until proton energies of @xmath48 gev . the associated determinations of @xmath23 and @xmath8 are presented in fig . [ fit - betabeams ] . the main conclusion is that setup ii is clearly better for the cp violating phase . not only the high energy setup provides a better precision , but it is able to resolve the degeneracies . in @xcite one may find the associated sensitivities of these setups for each parameter @xmath23 and @xmath8 . as a consequence , a @xmath49 effort to design beta beams for the upgraded cern sps ( @xmath50 gev ) appears justified . for the setups i ) and ii ) , as explained in the text.,width=245 ] electron capture is the process in which an atomic electron is captured by a proton of the nucleus leading to a nuclear state of the same mass number @xmath51 , replacing the proton by a neutron , and a neutrino . its probability amplitude is proportional to the atomic wavefunction at the origin , so that it becomes competitive with the nuclear @xmath1 decay at high @xmath52 . kinematically , it is a two body decay of the atomic ion into a nucleus and the neutrino , so that the neutrino energy is well defined and given by the difference between the initial and final atomic masses @xmath53 minus the excitation energy of the final nuclear state . in general , the high proton number @xmath52 nuclear beta - plus decay @xmath54 and electron - capture @xmath55 transitions are very `` forbidden '' , i.e. , disfavoured , because the energetic window open @xmath56 does not contain the important gamow - teller strength excitation seen in ( p , n ) reactions . there are a few cases , however , where the gamow - teller resonance can be populated ( see fig.[dy ] ) having the occasion of a direct study of the `` missing '' strength . for the rare - earth nuclei above @xmath43 , the filling of the intruder level @xmath57 for protons opens the possibility of a spin - isospin transition to the allowed level @xmath58 for neutrons , leading to a fast decay . our studies for neutrino beam preparation have used the @xmath59 ion with half live of 7.2 min , a branching ratio to neutrino channels of @xmath60 ( fully by ec ) and neutrino energy of @xmath61 mev in the c.m . frame as obtained from its decay to the single giant gamow - teller resonance in the daugther @xmath62 . decay of @xmath63.,width=245 ] a neutrino ( of energy @xmath64 ) that emerges from radioactive decay in an accelerator will be boosted in energy . at the experiment , the measured energy distribution as a function of angle @xmath65 and lorentz gamma @xmath66 of the ion at the moment of decay can be expressed as @xmath67 $ ] . the angle @xmath68 in the formula expresses the deviation between the actual neutrino detection and the ideal detector position in the prolongation of one of the long straight sections of the decay ring . the neutrinos are concentrated inside a narrow cone around the forward direction . if the ions are kept in the decay ring longer than the half - life , the energy distribution of the neutrino flux arriving to the detector in absence of neutrino oscillations is given by the master formula @xmath69 with a dilation factor @xmath70 . it is remarkable that the result is given only in terms of the branching ratio and the neutrino energy and independent of nuclear models . in eq . ( [ master ] ) , @xmath71 is the total number of ions decaying to neutrinos . for an optimum choice with @xmath72 around the first oscillation maximum , eq . ( [ master ] ) says that lower neutrino energies @xmath64 in the proper frame give higher neutrino fluxes . the number of events will increase with higher neutrino energies as the cross section increases with energy . to conclude , in the forward direction the neutrino energy is fixed by the boost @xmath73 , with the entire neutrino flux concentrated at this energy . as a result , such a facility will measure the neutrino oscillation parameters by changing the @xmath46 s of the decay ring ( energy dependent measurement ) and there is no need of energy reconstruction in the detector . for the study of the physics reach associated with such a facility , we combine two different energies for the same @xmath59 ion using two setups . in all cases we consider @xmath74 decaying ions / year , a water cerenkov detector with fiducial mass of @xmath75 kton and both appearance ( @xmath76 ) and disappearance ( @xmath77 ) events . setup i is associated with a five year run at @xmath9 ( close to the minimum energy to avoid atmospheric neutrino background ) plus a five year run at @xmath10 ( the maximum energy achievable at present sps ) , with a baseline @xmath4 km from cern to frejus . the results for setup i are going to be compared with those for setup ii , associated with a five year run at @xmath10 plus a five year run at @xmath11 ( the maximum achievable at the upgraded sps with proton energy of @xmath48 gev ) , with a baseline @xmath6 km from cern to canfranc . for the setup i we generate the statistical distribution of events from assumed values of @xmath23 and @xmath8 . the corresponding fit is shown in fig . [ fit - setupi ] for selected values of @xmath23 from @xmath78 to @xmath79 and covering a few values of the cp phase @xmath8 . as observed , the principle of an energy dependent measurement ( illustrated here with two energies ) is working to separate out the two parameters . with this configuration the precision obtainable for the mixing ( even at 1 degree ) is much better than that for the cp phase . from statistical distribution.,width=283 ] the corresponding exclusion plots which define the sensitivity to discover a non - vanishing mixing @xmath80 and cp violation @xmath81 are presented in fig . [ sens - theta ] and fig . [ sens - delta ] for varying confidence levels . for @xmath82 cl the sensitivity to a non - vanishing mixing is tipically around 1 degree . the corresponding sensitivity ( @xmath82 cl ) to see cp violation becomes significant for @xmath83 with values of the phase @xmath8 around @xmath84 to be distinguished from zero . sensitivity.,width=283 ] in the case of setup ii the longer baseline for @xmath10 leads to a value of @xmath41 well inside the second oscillation . in that case the associated strip in the ( @xmath23 , @xmath8 ) plane has a more pronounced curvature , so that the two parameters can be better disantangled . the statistical distribution generated for some assumed values of ( @xmath23 , @xmath8 ) has been fitted and the @xmath85 values obtained . the results are given in fig . [ fit - ec ] . qualitatively , one notices that the precision in the mixing is somewhat ( but no much ) better than that in setup i. on the contrary , the precision reachable for the cp phase is much better than that for setup i. one should emphasize that this improvement in the cp phase has been obtained with the neutrino channel only , using two appropriate different energies . the corresponding exclusion plots which define the sensitivities to discover a non - vanishing mixing @xmath80 and cp violation @xmath81 are presented in fig . [ sen - ec - theta ] and fig . [ sen - ec - delta ] for varying confidence levels . for @xmath82 cl the sensitivity to a non - vanishing mixing is , as before , significant up to around 1 degree . the corresponding sensitivity ( @xmath82 cl ) to see cp violation for @xmath83 becomes now significant with values of the phase @xmath8 around @xmath86 to be distinguished from zero . at the time of the operation of this proposed facility in setup ii it could happen that the connecting mixing @xmath23 is already known from the approved experiments for second generation neutrino oscillations , like double chooz , t2k and nova . to illustrate the gain obtainable in the sensitivity to discover cp violation from the previous knowledge of @xmath23 we present in fig . [ sen - delta-1par ] the expected sensitivity with the distribution of events depending on a single parameter @xmath8 for a fixed known value of @xmath23 . the result is impressive : even for a mixing angle of one degree , the cp violation sensitivity at @xmath87 cl reaches values around @xmath88 . the simulations of the physics output for both beta and ec beams indicate : \1 ) the upgrade to higher energy ( @xmath50 gev ) is crucial to have a better sensitivity to cp violation , which is the main objective of the next generation neutrino oscillation experiments , iff accompanied by a longer baseline . \2 ) the best @xmath41 in order to have a higher sensitivity to the mixing @xmath7 is not the same than that for the cp phase . like the phase - shifts , the presence of @xmath8 is easier to observe when the energy of the neutrino beams enters into the region of the second oscillation . the mixing is better seen around the first oscillation maximum , instead . in particular , setup ii in ec beams , i.e. , with @xmath89 between @xmath90 and @xmath75 and a baseline @xmath6 km ( canfranc ) , has an impressive sensitivity to cp violation , reaching precisions around @xmath86 , for @xmath82 cl , or better ( if some knowledge on the value of @xmath23 is already established ) . besides the feasibility studies for the machine , most important for physics is the study of the optimal configuration by combining low energy with high energy neutrino beams , short baseline with long baseline and/or ec monochromatic neutrinos with @xmath26 @xmath2 antineutrinos . among the possible systematics associated with the proposed experiments one should define a program to determine independently the relevant cross sections of electron and muon neutrinos and antineutrinos with water in the relevant energy region from several hundreds of mev s to 1 gev or so . the result of the synergy of neutrino physics with nuclear physics ( eurisol ) and lhc physics ( sps upgrade ) for the facility at cern could be completed with the synergy with astroparticle physics for the detector , which could be common to neutrino oscillation studies with terrestrial beams , atmospheric neutrinos ( sensitive to the neutrino mass hierarchy through matter effects @xcite ) , supernova neutrinos and proton decay . for the extensive literature see for example the reviews m. apollonio , _ et al _ , [ hep - ex/0210192 ] ; j. j. gmez - cadenas and d. a. harris , ann . nucl . part . * 52 * ( 2002 ) 253 and the annual proceedings of the international nufact workshop . h. minakata and h. nunokawa , phys . b * 495 * ( 2000 ) 369 [ arxiv : hep - ph/0004114 ] . j. j. gmez - cadenas _ et al . _ [ cern working group on super beams collaboration ] , [ hep - ph/0105297 ] . h. minakata and h. nunokawa , jhep * 0110 * ( 2001 ) 001 , [ hep - ph/0108085 ] . p. huber , m. lindner and w. winter , nucl . b * 645 * ( 2002 ) 3 [ hep - ph/0204352 ] . g. barenboim _ et al . _ , [ hep - ex/0206025 ] . v. martemianov _ et al _ , phys . 66 ( 2003 ) 1934 - 1939 ; yad.fiz . 66 ( 2003 ) 1982 - 1987 [ hep - ex/0211070 ] ; h. minakata _ et al _ , phys . d * 68 * ( 2003 ) 033017 [ erratum - ibid . d * 70 * ( 2004 ) 059901 ] [ hep - ph/0211111 ] . phys . lett . b * 580 * ( 2004 ) 216 [ hep - ph/0309323 ] . p. huber , m. lindner , t. schwetz and w. winter , nucl . b * 665 * ( 2003 ) 487 [ hep - ph/0303232 ] . f. ardellier _ et al . _ , [ hep - ex/0405032 ] . p. zucchelli , phys . lett . * b532 * ( 2002 ) 166 . j. burguet - castell , d. casper , e. couce , j. j. gomez - cadenas and p. hernandez , nucl . b * 725 * , 306 ( 2005 ) , [ hep - ph/0503021 ] . a. cervera _ et al_. , nucl . b * 579 * ( 2000 ) 17 ; erratum - ibid b * 593 * ( 2001 ) 731 . m. c. gonzalez - garcia , global analysis of neutrino , in nobel symposium 2004 : neutrino physics , haga slott , enkoping , sweden , 19 - 24 aug 2004 , [ hep - ph/0410030 ] . j. bernabeu , nucl . b * 49 * ( 1972 ) 186 . j. sato , phys . lett . * 95 * 131804 ( 2005 ) , [ hep - ph/0503144 ] . m. c. banuls , g. barenboim and j. bernabeu , phys . b * 513 * , 391 ( 2001 ) [ arxiv : hep - ph/0102184 ] ; j. bernabeu , s. palomares - ruiz , a. perez and s. t. petcov , phys . b * 531 * , 90 ( 2002 ) ; j. bernabeu , s. palomares ruiz and s. t. petcov , nucl . b * 669 * , 255 ( 2003 ) .	the goal for future neutrino facilities is the determination of the @xmath0 $ ] mixing and cp violation in neutrino oscillations . this will require precision experiments with a very intense neutrino source . with this objective the creation of neutrino beams from the radioactive decay of boosted ions by the sps of cern from either beta or electron capture transitions has been propossed . we discuss the capabilities of such facilities as a function of the energy of the boost and the baseline for the detector . we conclude that the sps upgrade to 1000 gev is crucial to have a better sensitivity to cp violation if it is accompanied by a longer baseline . we compare the physics potential for two different configurations . in the case of beta beams , with the same boost for both @xmath1 ( neutrinos ) and @xmath2 ( antineutrinos ) , the two setups are : i ) @xmath3 , @xmath4 km ( frejus ) ; ii ) @xmath5 , @xmath6 km ( canfranc ) . in the case of monochromatic ec beams we exploit the energy dependence of neutrino oscillations to separate out the two parameters @xmath7 and the cp phase @xmath8 . setup i runs at @xmath9 and @xmath10 ( maximum achievable at present sps ) to frejus , whereas setup ii runs at @xmath10 and @xmath11 ( maximum achievable at upgraded sps ) to canfranc . the main conclusion is that , whereas the gain in the determination of @xmath7 is rather modest , setup ii provides much better sensitivity to cp violation .
the global analysis of nuclear parton distributions ( npdfs ) is driven by the experimental fact that the deep inelastic structure functions @xmath1 measured from nuclear targets show a significant deviation from the free proton ones @xcite . perhaps the most simple theoretical approach to this observation is to make use of the factorization theorem of qcd that has proven to provide excellent description of inclusive cross - sections in free - nucleon collisions . in this approach , the cross - sections are of the generic form @xmath2 where where @xmath3 is the perturbative qcd ( pqcd ) matrix element squared and @xmath4s are the non - perturbative parton densities whose scale evolution obeys the dglap equations @xcite . the purpose of the global analysis of npdfs is to find out whether the observed differences in the structure functions , the nuclear modifications , can consistently be absorbed in to the input parton densities in other words , do the nuclear modifications effectively factorize . if they do , the resulting npdfs are of great practical interest , since they can be used as a input in any process that can be factorized as in eq . ( [ eq : facqcd ] ) . in this framework the deep question about the dynamical origin of nuclear modifications is not addressed , on the contrary , one must be as unbiased to any model as possible . indeed , three independent groups have shown that this approach works quite well : * eks98 @xcite was the first global analysis demonstrating that using the leading - order ( lo ) pqcd formalism , requiring momentum and baryon number conservation , one can reproduce the data from measurements of deep inelastic lepton - nucleus scattering ( dis ) and drell - yan dilepton production ( dy ) in proton - nucleus collisions . the fit was done only by eye . * hkm @xcite & hkn @xcite which are lo qcd analyses as well , were the first ones to exploit the automated @xmath0-minimization procedure and to estimate npdf uncertainties . * nds @xcite brought the global analysis of npdfs to the next - to - leading order level in pqcd . based on our earlier work , eks98 , we have performed a global reanalysis of npdfs published in @xcite and reported here . what was new compared to the eks98 were the automated @xmath0-minimization procedure and a first attemp for estimating the uncertainties . although our main objective was to see whether we can improve the eks98-fit and study the uncertainties , this was anyway a necessary stepping stone for us before we can extend our analysis to nlo - level . motivated by the brahms data @xcite on inclusive hadron production in d+au collision , which show a systematic suppression relative to p+p at forward rapidities , we raise an intriguing question about the possibility of having clearly stronger gluon shadowing than what has been hitherto seen in the global dglap analyses . we define the pdfs @xmath5 of bound protons in a nucleus with mass number @xmath6 as @xmath7 where @xmath8 refers to the latest free proton pdfs by the cteq collaboration @xcite . for the bound neutrons we assume the isospin symmetry @xmath9 and vice versa . what we actually parametrize , are the nuclear modifications @xmath10 at the initial scale @xmath11 . at present , the lack of precision data forces us to consider only three different modifications : @xmath12 for all valence quarks , @xmath13 for all sea quarks , and @xmath14 for gluons . the fit functions are parametrized in three pieces ( c.f . fig . [ ini ] ) : @xmath15 , \qquad \qquad x\le x_a^a \nonumber \\ & & r_2^a(x ) = a_0^a + a_1^a x + a_2^a x^2 + a_3^a x^3 , \qquad \qquad \qquad \quad \ , \ , \ ; \ ; \ : \qquad x_a^a \le x \le x_e^a\\ \label{r2 } & & r_3^a(x ) = \frac{b_0^a - b_1^a x}{(1-x)^{\beta^a } } , \qquad \qquad \qquad \qquad \qquad \quad \ : \ : \qquad \qquad \qquad x_e^a \le x \nonumber,\end{aligned}\ ] ] the first one covers the region from shadowing to anti - shadowing maximum at @xmath16 , the second comes down to emc - minimum at @xmath17 , and the third is for the fermi - motion part . the @xmath6-dependence of the fit parameters is assumed to follow a power law @xmath18 where we have chosen carbon ( @xmath19 ) as a reference nucleus . after fixing the continuity of @xmath20s and their first derivatives at @xmath16 and @xmath17 , one is still left with @xmath21 free parameters of which the baryon number and momentum conservation eat only @xmath22 away . this was still too much in order to obtain converging fits , and lots of manual work was needed too see which parameters were the most relevant ones . at the end there was @xmath23 fit parameters . the experimental input in our analysis was about @xmath24 points of dis- and dy - data in a form @xmath25 these covered @xmath26 different nuclei from helium up to lead . [ fig : data1 ] and [ fig : data2 ] show some of these data and comparison to our result from minimization of @xmath27 calculated @xmath28 ( filled symbols ) are compared to slac @xcite , e665 @xcite , nmc 95 @xcite and reanalysed nmc 95 data @xcite . the asterisks denote our results calculated at the initial scale @xmath29 , these are for the smallest-@xmath30 data points whose scales lie in the region @xmath31 . ] calculated @xmath28 ( filled symbols ) are compared to slac @xcite , e665 @xcite , nmc 95 @xcite and reanalysed nmc 95 data @xcite . the asterisks denote our results calculated at the initial scale @xmath29 , these are for the smallest-@xmath30 data points whose scales lie in the region @xmath31 . ] calculated @xmath32 and @xmath33 ( filled symbols ) are compared to e772 @xcite and nmc data @xcite ] calculated @xmath32 and @xmath33 ( filled symbols ) are compared to e772 @xcite and nmc data @xcite ] the number that characterizes the goodness of the fit is @xmath34 ( @xmath35 ) which should be less than @xmath36 if the fit is any good . in our case @xmath37 indicating that the theory fits the data very well and that there is no serious sign why pqcd could not be trusted within the considered kinematical range also in the nuclear environment . the obtained nuclear modifications at the initial scale for lead are shown in fig . this figure also presents our uncertainty estimates based on hessian method of quantifying the uncertainties @xcite , in which one expands the @xmath0 around the minimum w.r.t fit parameters @xmath38 as @xmath39 and the uncertainty of any quantity @xmath40 depending on pdfs is then obtained from @xmath41 ^ 2 = \delta\chi^2 \sum_{i , j } \left(\frac{\partial f(\hat\xi)}{\partial \xi_i}\right ) h_{ij}^{-1}\left(\frac{\partial f(\hat\xi)}{\partial \xi_j}\right ) . \label{errors}\ ] ] for an ideal @xmath0-distribution @xmath42 , which we used too , corresponds to `` one sigma''-error . due to technical difficulties obtaining converging fits , the emc minimum of gluons and sea quarks was fixed to follow the valence quarks at @xmath43 . this resulted as an unreliably small error bands for gluons and sea quarks , and they had to be computed separately . they are the `` large-@xmath30 errors '' in fig . the combined uncertainties are shown as a yellow bands . interestingly , the old eks98 parametrization lies within these uncertainties and there is no reason to release a new parametrization eks98 works just fine . however , one should be very cautious about these error bands ! first , below @xmath44 there is no experimental data above @xmath45 , and the behaviour at small-@xmath30 region is constrained only by the sum rules and is bound to the form of the fit function . second , the pdfs themselves depend on choices and conventions , like kinematical cuts , choosing the factorization scale , treatment of heavy quarks , and choosing and weighting data sets in forming @xmath0 . how these choices affect the obtained pdfs is not seen in the error analysis performed here the error bands only reflect the experimental data and their errors . third , there is no well - established way to choose @xmath46 , and our choice @xmath42 is probably quite restrictive . the dy and dis data leave the gluons very unconstrained gluon dependence comes only through the dglap - evolution . one possible way to constrain the gluon sector could be the inclusive hadron production at rhic . figure [ strong ] shows data from the brahms collaboration @xcite for @xmath47 , the ratio between charged hadron production in d+au and p+p collisions as a function of hadron s @xmath48 . minimum bias inclusive hadron production cross sections in d+au collisions divided by that in p+p collisions at @xmath49 gev at rhic . the ratio @xmath50 is shown as a function of hadrons transverse momentum at four different pseudorapidities . the brahms data @xcite are shown with the statistical error bars and the shaded systematic error limits . a pqcd calculation for @xmath51 production with the nuclear modifications from present work and kkp fragmentation functions is shown by the black lines , and that with the strong gluon shadowing , shown at right , by green lines . ] minimum bias inclusive hadron production cross sections in d+au collisions divided by that in p+p collisions at @xmath49 gev at rhic . the ratio @xmath50 is shown as a function of hadrons transverse momentum at four different pseudorapidities . the brahms data @xcite are shown with the statistical error bars and the shaded systematic error limits . a pqcd calculation for @xmath51 production with the nuclear modifications from present work and kkp fragmentation functions is shown by the black lines , and that with the strong gluon shadowing , shown at right , by green lines . ] from the point - of - view of pqcd , this is computed via @xmath52 where the @xmath53s are the fragmentation functions . in fig . [ strong ] the pqcd calculation which employs the npdfs from the present analysis and kkp @xcite fragmentation functions , is shown by dark line . it is evident that , especially at very forward direction , our prediction is above the data and the shape is not well reproduced . since hadron production at the kinematical corner of low-@xmath48 and forward rapidity reaches the small-@xmath30 region of pdfs where the gluon distributions are the dominant ones , this could be signaling a larger uncertainty in our gluon modifications than seen in fig . for this reason we present an example of gluon shadowing that is much stronger . this is shown on the right hand side of fig . [ strong ] , and it really helps : using this gluon modification the corresponding curve for @xmath47 is brought clearly closer to the brahms data . however , it s still too early to draw very strong conclusions from this observation , but a systematic study in the context of global dglap analysis is needed @xcite . one should also bear in mind that here we are considering effects at very low - pt region , where the simple lo pqcd picture is pushed to its very limits although one can argue that in ratios like @xmath54 some higher order effects would partially cancel . anyway , one should be very careful when interpreting these results . for example , it has been conjectured , that this particular brahms data set could be a sign of parton saturation at work @xcite . cas is supported by the 6th framework programme of the european community under the marie curie contract meif - ct-2005 - 024624 . we thank the academy of finland , project 115262 , and the national graduate school of particle and nuclear physics in finland for financial support . 99 m. arneodo , phys . * 240 * ( 1994 ) 301 . n. armesto , j. phys . g * 32 * ( 2006 ) r367 [ arxiv : hep - ph/0604108 ] . y. l. dokshitzer , perturbation theory in quantum sov . jetp * 46 * ( 1977 ) 641 [ zh . * 73 * ( 1977 ) 1216 ] ; v. n. gribov and l. n. lipatov , yad . fiz . * 15 * ( 1972 ) 781 [ sov . j. nucl . phys . * 15 * ( 1972 ) 438 ] ; v. n. gribov and l. n. lipatov , yad . fiz . * 15 * ( 1972 ) 1218 [ sov . j. nucl . phys . * 15 * ( 1972 ) 675 ] ; g. altarelli and g. parisi , nucl . b * 126 * ( 1977 ) 298 . k. j. eskola , v. j. kolhinen and p. v. ruuskanen , nucl . b * 535 * ( 1998 ) 351 [ arxiv : hep - ph/9802350 ] . k. j. eskola , v. j. kolhinen and c. a. salgado , eur . j. c * 9 * ( 1999 ) 61 [ arxiv : hep - ph/9807297 ] . m. hirai , s. kumano and m. miyama , phys . d * 64 * ( 2001 ) 034003 [ arxiv : hep - ph/0103208 ] . m. hirai , s. kumano and t. h. nagai , phys . c * 70 * ( 2004 ) 044905 [ arxiv : hep - ph/0404093 ] . d. de florian and r. sassot , phys . d * 69 * ( 2004 ) 074028 [ arxiv : hep - ph/0311227 ] . k. j. eskola , v. j. kolhinen , h. paukkunen and c. a. salgado , jhep * 0705 * ( 2007 ) 002 [ arxiv : hep - ph/0703104 ] . i. arsene _ et al . _ [ brahms collaboration ] , phys . * 93 * ( 2004 ) 242303 [ arxiv : nucl - ex/0403005 ] . j. pumplin , d. r. stump , j. huston , h. l. lai , p. nadolsky and w. k. tung , jhep * 0207 * ( 2002 ) 012 [ arxiv : hep - ph/0201195 ] . m. hirai , s. kumano and n. saito [ asymmetry analysis collaboration ] , phys . d * 69 * ( 2004 ) 054021 [ arxiv : hep - ph/0312112 ] . d. m. alde _ et al . _ , phys . lett . * 64 * ( 1990 ) 2479 . j. gomez _ et al . _ , phys . d * 49 * ( 1994 ) 4348 . m. r. adams _ et al . _ [ e665 collaboration ] , z. phys . c * 67 * ( 1995 ) 403 [ arxiv : hep - ex/9505006 ] . p. amaudruz _ et al . _ [ new muon collaboration ] , nucl . b * 441 * ( 1995 ) 3 [ arxiv : hep - ph/9503291 ] . m. arneodo _ et al . _ [ new muon collaboration . ] , nucl . b * 441 * ( 1995 ) 12 [ arxiv : hep - ex/9504002 ] . m. arneodo _ et al . _ [ new muon collaboration ] , nucl . b * 481 * ( 1996 ) 23 . b. a. kniehl , g. kramer and b. potter , nucl . b * 582 * ( 2000 ) 514 [ arxiv : hep - ph/0010289 ] . k. j. eskola , h. paukkunen and c. a. salgado , in preparation . r. baier , a. kovner and u. a. wiedemann , phys . d * 68 * , 054009 ( 2003 ) ; d. kharzeev , y. v. kovchegov and k. tuchin , phys . d * 68 * ( 2003 ) 094013 ; j. l. albacete , n. armesto , a. kovner , c. a. salgado and u. a. wiedemann , phys . rev . lett . * 92 * , 082001 ( 2004 ) .	in this talk , we shortly report results from our recent global dglap analysis of nuclear parton distributions . this is an extension of our former eks98-analysis improved with an automated @xmath0 minimization procedure and uncertainty estimates . although our new analysis show no significant deviation from eks98 , a sign of a significantly stronger gluon shadowing could be seen in the rhic brahms data .
over the last several years , many theoretical and experimental efforts have been devoted for investigations of deeply - bound kaonic nuclei @xcite , but there is no conclusive proof on their existence at the present stage . in order to study properties of antikaon @xmath12 in nuclear medium , it is an important subject to verify the reliable evidence of deeply - bound kaonic nuclei because antikaons might condense in the interior of compact star @xcite . among various antikaon nuclear systems , the three - body @xmath13 bound state with a configuration of @xmath14_{i=1/2}$ ] is expected to be the lightest and the most fundamental kaonic nucleus . this state is symbolically called `` @xmath0 '' in this paper . in 1963 , nogami @xcite has firstly suggested a possible existence of the @xmath0 bound state by a rather crude calculation . about 40 years later , yamazaki and akaishi @xcite restarted to study the structure of the @xmath0 bound state on the basis of a quantitative few - body calculation with the brueckner theory , and predicted that the binding energy and width for @xmath2 are @xmath15 mev and @xmath16 mev , respectively . shevchenko , gal and mare@xmath17 @xcite performed the @xmath18-@xmath19 coupled channel calculation for the @xmath0 bound state by the faddeev approach , and obtained @xmath20 = 55 - 70 mev and @xmath21 = 95 - 110 mev . several theoretical works @xcite also supported the existence of the @xmath0 bound state , but the calculated binding energy and width are not converged theoretically . on the other hand , finuda collaboration at da@xmath22ne @xcite reported experimentally evidence of a deeply - bound @xmath0 state in invariant mass spectroscopy by @xmath0 @xmath23 @xmath24 + @xmath25 decay processes from @xmath2 absorption on @xmath26li , @xmath27li and @xmath28c at rest . the measured energy and width are @xmath29 mev and @xmath30 mev , respectively . however , magas @xmath31 @xmath32 @xcite claimed that it is not necessary to postulate a @xmath0 bound state in order to explain a bump structure in the finuda data , because of the @xmath33+@xmath34 pair interaction with the residual nucleus , following the @xmath2 absorption on two nucleons in these nuclei . although this data analysis is continued by the finuda group , the situation is still unsettled . more experimental data are required in order to confirm whether the @xmath0 system has a deeply - bound state or not . recently , iwasaki @xmath31 @xmath32 @xcite proposed a new experiment searching the deeply - bound @xmath0 state at j - parc , by the missing - mass spectrum of the @xmath1he(in - flight @xmath2,@xmath3 ) reaction , together with the invariant - mass spectra detecting all particles via decay processes from the @xmath0 bound state ( j - parc e15 ) . our purpose is to clarify theoretically the expected inclusive and semi - exclusive spectra of the @xmath1he(in - flight @xmath2,@xmath3 ) reaction for preparing the forthcoming j - parc e15 experiment . as for ( in - flight @xmath2,@xmath34 ) reactions on @xmath28c and @xmath35o targets , kishimoto @xmath31 @xmath36 . @xcite performed measurements for searching deeply - bound kaonic nuclei , but no clear evidence was observed . @xmath32 @xcite calculated the corresponding inclusive spectra by @xmath28c,@xmath35o(in - flight @xmath2 , @xmath25 ) reactions , using @xmath2 optical potentials based on chiral unitary and phenomenological approaches ; they suggested that any distinct peaks of deeply - bound kaonic states will not be observed even if the bound states exist . the predicted states can not be separated due to their broad width . it would be naively expected that if we choose a lighter nucleus as a target , a signal for formation of kaonic nuclei appears more clearly , whereas no theoretical calculation has been done for the @xmath1he target so far . yamazaki and akaishi @xcite also proposed the @xmath0 formation via @xmath37 , assuming a @xmath38 doorway . such a @xmath0 experiment with a @xmath394 gev proton beam on a deuterium target is planned by fopi collaboration at gsi @xcite . in this letter , we demonstrate firstly the inclusive and semi - exclusive spectra of the @xmath1he(in - flight @xmath2 , @xmath3 ) reaction at @xmath4 = 1.0 gev / c and @xmath5 within the distorted wave impulse approximation ( dwia ) , in order to predict the production cross section of the @xmath0 bound state and to examine the experimental feasibility . as studied in ref . @xcite , we employ the green s function method by morimatsu and yazaki @xcite in the dwia framework , which describes well unstable hadron nuclear systems , comparing theoretical spectra with experimental ones @xcite . in the dwia framework @xcite , the inclusive double - differential cross section of the @xmath1he(in - flight @xmath2,@xmath3 ) reaction at the forward direction @xmath40 in the lab system is written @xcite as @xmath41^{({\rm elem } ) } \ , s(e ) , \label{eq : dwia}\end{aligned}\ ] ] where @xmath42^{({\rm elem})}$ ] is a fermi - averaged cross section for the elementary @xmath43 forward scattering which is equivalent to a backward @xmath44 elastic scattering @xcite in the lab system . here we consider only the non - spin - flip @xmath45 reaction channel . the kinematical factor @xmath46 @xcite expresses the translation from the two - body @xmath2-@xmath34 lab system to the @xmath2-@xmath1he lab system @xcite , and it is defined as @xmath47 where @xmath4 is the incident @xmath48 momentum , and @xmath49 and @xmath50 ( @xmath51 and @xmath52 ) denote the outgoing neutron momenta ( energies ) for a nucleon target and @xmath1he target , respectively , in the lab frame ; @xmath53 is the residual @xmath2 energy in the final state . @xmath54 denotes the strength function of the @xmath0 system , which is expressed in the green s function method @xcite , as @xmath55 \label{eq : s(e)}\end{aligned}\ ] ] with @xmath56 where @xmath57 ( @xmath58 ) is distorted waves of the incoming @xmath2 ( outgoing neutron ) with the momentum @xmath59 ( @xmath60 ) ; @xmath61 , @xmath62 and @xmath63 are the masses of the @xmath1he target , the core - nucleus and the @xmath0 system , respectively . the factor of @xmath64 and @xmath65 takes into account the recoil effects . here we assumed to use the recoil factor of @xmath65 in not only @xmath58 but also @xmath57 , for simplicity . @xmath66 is a hole - state wave function for a struck neutron in the target , where @xmath67 denotes the complete set of eigenstates for the system . we employ a simple ( 1s)@xmath1 harmonic oscillator model for the @xmath1he wave function . then , the relative @xmath68-@xmath34 wave function has the form of @xmath69 , @xmath70 . the size parameter @xmath71 is taken to be 1.30 fm , which reproduces the recent experimental r.m.s charge radius of @xmath1he , @xmath72 fm @xcite . in this calculation , we assumed the `` @xmath6 '' pair as a rigid core with @xmath73 , for simplicity . this assumption would be suitable for describing the considerable structure of the deeply - bound @xmath0 state , as mentioned below . therefore , green s function for the @xmath0 system , @xmath74 , in eq.([eq : s(e ) ] ) is obtained by solving numerically the following klein - gordon equation with a @xmath2-@xmath6 optical potential @xmath75 ; @xmath76 where @xmath77 is the reduced mass between the @xmath2 and the `` @xmath6 '' core - nucleus , and @xmath78 the coulomb potential with the finite nuclear size effect . moreover , the strength function @xmath54 can be decomposed into two parts @xcite as @xmath79 where @xmath80 describes the @xmath2 conversion processes including the decay modes of @xmath81 ( mesonic ) from the one - body absorption and @xmath82 ( nonmesonic ) from the two - body absorption , and @xmath83(e ) the @xmath2 escaping processes leaving the core - nucleus in the ground state : @xmath84 where @xmath85 is the free green s function . here we used an abbreviated notation for @xmath86 , @xmath87 and @xmath75 . we evaluate the distorted waves within the eikonal approximation as @xmath88 , % \nonumber \label{eq : eikonalwave- } \\ \chi^{(+ ) } \left ( \mbox{\boldmath{$p$}}_{k^- } , \mbox{\boldmath{$r$ } } \right ) & = & \exp \left [ + i \mbox{\boldmath{$p$}}_{k^-}\cdot\mbox{\boldmath{$r$ } } - \frac{i}{v_{k^- } } \int^z_{-\infty } u_{k^-}(\mbox{\boldmath{$b$}},z')dz ' \right ] \label{eq : eikonalwave+}\end{aligned}\ ] ] with an impact parameter coordinate @xmath89 and the optical potential for @xmath90 or @xmath3 , @xmath91 where @xmath92 is a nuclear density distribution , and @xmath93 and @xmath94 are the isospin - averaged total cross section and the ratio of the real to imaginary parts of the forward amplitude for the @xmath95 scattering , respectively . these parameters are chosen to be @xmath96 mb and @xmath97 @xcite . yamazaki and akaishi @xcite obtained the @xmath0 optical potential between the @xmath2 and the `` @xmath6 '' pair , performing a variational three - body calculation . this ya optical potential is parametrized in a gaussian form with a range - parameter @xmath98 fm , as @xmath99 , \label{eq : kppoptpot}\end{aligned}\ ] ] where @xmath100 mev and @xmath101 mev . note that these parameters are determined within a non - relativistic framework . solving the klein - gordon equation with this potential , we find that the @xmath2 binding energy and width account for @xmath102 mev and @xmath103 mev , respectively . although these values are slightly changed from @xmath15 mev and @xmath16 mev given in ref . @xcite , we employ the ya potential having ( @xmath7 , @xmath8 ) = ( @xmath104 mev , @xmath105 mev ) as a standard one , for a first step toward our quantitative investigation . shevchenko , gal and mare@xmath17 @xcite also obtained @xmath106 mev and @xmath107 mev as the maximum values for the @xmath0 bound state by the faddeev calculation . we can simulate @xmath108 mev and @xmath109 mev as a complex klein - gordon energy , adjusting the parameters to ( @xmath7 , @xmath8 ) = ( @xmath110 mev , @xmath111 mev ) within the potential form of eq.([eq : kppoptpot ] ) . we adopt these values of @xmath112 as an upper limit for the reliable parameters . let us consider the @xmath1he(in - flight @xmath2,@xmath3 ) reaction at the incident @xmath2 momentum @xmath4=1.0 gev / c and the forward direction @xmath113 , where the lab cross section of the elementary reaction process is maximal @xcite . the lab cross section for the @xmath45 forward scattering is @xmath3927 mb / sr @xcite in free space , and is reduced to @xmath3915 mb / sr with fermi - averaging @xcite . these values are about three times larger than those for the @xmath114 forward scattering for the ( @xmath2 , @xmath25 ) reaction at @xmath115 1.0 gev / c , e.g. , 8.8 mb / sr @xcite in free space and 5.2 mb / sr @xcite with fermi - averaging . we adopt 15 mb / sr as a value of @xmath42^{({\rm elem})}$ ] in eq.([eq : dwia ] ) , and we notice that the momentum transfer @xmath116 is negative ( @xmath117 ) which means that the residual @xmath2 recoils backwards relative to the incident @xmath2 in the lab frame , therefore , the kinematical factor @xmath46 in eq.([eq : beta ] ) enhances the spectrum by a factor 1 - 2 depending on @xmath118 1.0 - 1.4 gev / c . for missing - mass spectroscopy , fig . [ yaspec ] shows the calculated inclusive spectrum of the @xmath1he(in - flight @xmath2,@xmath3 ) reaction as a function of the ejected neutron momentum @xmath50 , and the components of the partial - wave angular momentum states with @xmath119 . here we used the ya potential with ( @xmath7 , @xmath8)=(@xmath104 mev , @xmath105 mev ) . we find that a distinct peak of the @xmath0 bound state with @xmath9 , @xmath120=0@xmath121 ( @xmath122 ) appears around @xmath123 1280 mev / c , which has @xmath124 50 mev and @xmath125 70 mev . the absolute value of the integrated formation cross section amounts to 3.5 mb / sr . note that the effective neutron number with the distortion for the @xmath0 bound state is @xmath126= 0.23 with @xmath127= 0.47 , of which value is more than ten times as large as the effective proton number @xmath128= 0.013 with @xmath127=0.095 for the @xmath2 nuclear @xmath129 bound state via the @xmath130 transition in @xmath28c @xcite . this is a major advantage of the use of the @xmath131-shell target nucleus such as @xmath1he . we stress that the calculated spectrum consists of mainly the states with @xmath132 below the @xmath2 emitted threshold ( @xmath118 1224 mev / c ) , whereas the components of @xmath133 dominates continuum states for the quasi - free ( qf ) region . this @xmath134 dominance in the bound region is caused by a small size of the @xmath131-shell nucleus , contrast to the case of @xmath25-shell nuclei , @xmath28c and @xmath35o . moreover , the recoil effects for the light @xmath131-shell nuclear target play an important role in increasing the cross section of the @xmath0 bound state ; if we replace the recoil factors of both @xmath65 and @xmath64 by 1 in eq . ( [ eq : chi+- ] ) , the cross section of the peak in the bound region is reduced by half , whereas the yields in the qf region grow up , and the qf peak is shifted to the lower @xmath50 side and its width is broader . when we use a reasonable factor of @xmath135 where @xmath136 is the mean mass of @xmath61 and @xmath63 , instead of @xmath65 , we find that the cross section around the peak in the bound region is enhanced by about 10% . fig.1 in fig . [ yaspec2 ] , we display contributions of the @xmath2 escape and conversion components into the inclusive spectrum , using the ya potential . since the `` @xmath6 '' rigid - core nucleus is assumed in this calculation , the @xmath18 continuum spectrum is described as @xmath2+``@xmath6 '' continuum states . thus the @xmath2 escape spectrum in eq.([eq : sesc ] ) might not be fully explained in the qf region where the @xmath1he break - up processes of @xmath137 occur . on the other hand , this assumption seems to work well in the bound region ; decay processes of @xmath138 and @xmath139 are effectively described by the imaginary potential , im@xmath140 , if we choose an appropriate one . such a conversion spectrum in eq.([eq : scon ] ) expresses semi - exclusive @xmath141 processes detecting the selected particles in invariant - mass spectroscopy at the j - parc e15 experiment . therefore , the treatment of the `` @xmath6 '' core - nucleus would be justified in not only the bound region but also the continuum region , as long as we concern the conversion spectrum . in fig . [ sgmspec2 ] , we show also the escape and conversion spectra using ( @xmath7 , @xmath8)=(@xmath110 mev , @xmath111 mev ) , in order to simulate the results by shevchenko , gal and mare@xmath17 @xcite . we confirm that a peak of the @xmath0 bound state is broadened because a value of @xmath142 is so larger than that obtained by the ya potential . the corresponding bump is barely recognized in the spectrum . fig.2 fig.3 mare@xmath17 , friedman and gal @xcite introduced a phase space suppression factor @xmath143 multiplying the imaginary part of the @xmath2 optical potential . when a pole of the @xmath0 bound state is located below the @xmath19 threshold in the complex energy plane , its width would be narrower by the phase space suppression because the @xmath144 decay channel is closed there . several predictions for the @xmath0 bound state @xcite suggest that its pole resides above the @xmath19 threshold , except the results in ref . @xcite . in order to see the effects , we replace @xmath8 by @xmath145 in eq.([eq : kppoptpot ] ) . here we assumed that the @xmath2 is absorbed from 80 % one - nucleon and 20 % two - nucleons , as shown in refs . @xcite . indeed , the helium bubble chamber experiment @xcite suggested that the ratio of @xmath2 absorption on two nucleons to all @xmath2 absorption process in @xmath146he at rest amounts to 16 % . taking into account the phase space factor , we adjust a value of @xmath8 to be @xmath147 mev so as to reproduce the same @xmath142 of the @xmath0 bound state . since @xmath80 can be further decomposed into the corresponding spectra from the one- and two - nucleon absorptions , we can estimate each component for these decay modes . [ ay - edep ] shows the effects of the phase space factor in the calculated spectrum . we find that the shape of the inclusive spectrum is not so changed when the phase space factor is turned on ; the spectrum of the @xmath0 bound state , however , is sizably reduced near the @xmath19 threshold ( @xmath148 mev / c ) because the @xmath0 below the threshold can decay into only the @xmath149 process from the two - nucleon absorption , which contributes slightly to the spectrum on the whole . fig.4 it is worth studying shape behaviors of the strength function @xmath54 for the @xmath0 bound state on the parameters of @xmath7 and @xmath8 , systematically . as shown in fig . [ svdep ] , we change the value of @xmath7 from @xmath150 mev to @xmath110 mev , keeping @xmath101 mev . we see that a peak of the bound state gives rise to a cusp - like structure at @xmath151-@xmath152 mev ( shallow ) . in fig . [ swdep ] , we show the shape behavior of @xmath54 when we change the value of @xmath8 from @xmath153 mev to @xmath154 mev , keeping @xmath100 mev . we find that choosing @xmath155 mev ( deep ) , the peak is completely invisible even if the bound state exists . consequently , in order to identify a peak of the @xmath0 bound state in the inclusive spectrum , we realize that the value of @xmath7 must be deeper than @xmath156 mev and also that of @xmath8 shallower than @xmath156 mev . the parameter set for the ya potential @xcite fulfills this condition , while that simulating the results by shevchenko , gal and mare@xmath17 @xcite stands on the edge of the signal observation in terms of the width . we believe that it is possible to extract variable information on the @xmath2 optical potential from the shape behavior of the inclusive and semi - exclusive spectra . fig.5 fig.6 dot@xmath157 @xmath31 @xmath32 @xcite predicted the possibility of extreme nuclear shrinkage in the @xmath158 system ( strange tribaryon ) by anti - symmetrized molecular dynamics ( amd ) calculations . such nuclear contraction is strongly related to the antikaon condensation in compact star @xcite . it is interesting to evaluate the nuclear contraction in the @xmath0 bound state by the ( @xmath2 , @xmath3 ) reaction . when a harmonic oscillator model is applied to @xmath1he , the r.m.s . distance of the `` @xmath6 '' pair is 1.90 fm in the @xmath0 bound state , whose value is reduced from that of 2.25 fm in @xmath1he , as discussed in ref . however , this effect is not expected to be observed because the cross section of the @xmath0 bound state is only slightly suppressed by about 4% . our simplified calculation in this paper would bring some uncertainties for the resultant spectrum : only the @xmath159 forward scattering has been considered , omitting the @xmath160 charge exchange process which can also contribute to the @xmath0 formation through the coupling between @xmath0 and @xmath161 channels . this contribution is roughly estimated by replacing the fermi - averaged transition amplitude @xmath162 for the elementary @xmath45 process by the isoscalar transition amplitude @xmath163 which is equivalent to a @xmath164-channel isospin-0 one for ( @xmath2,@xmath165 reactions , including a spectroscopic factor for the @xmath0 @xmath166=1/2 state formed on @xmath1he . as a result , the cross section of the @xmath0 bound state is enhanced by about 18 % with fermi - averaging . it is noticed that the @xmath0 state with @xmath166=1/2 dominates the ( @xmath2,@xmath3 ) reaction at @xmath4= 1.0 gev / c because @xmath167 , where @xmath168 is the isovector transition amplitude @xcite . this nature justifies our assumption that we treat @xmath14_{i=1/2}$ ] as @xmath0 restrictedly in this calculation . in order to get more quantitative results , a full coupled - channel calculation would be needed . moreover , the choice of the parameters in the eikonal distorted waves also changes the absolute value of the cross section , but the distortion effect is not significant for @xmath1he . for the in - flight ( @xmath2,@xmath34 ) reaction , it would be not appropriate to use the decay rate measured by @xmath2 absorption at rest , considering that its value depends on atomic orbits where @xmath2 is absorbed through atomic cascade processes @xcite . more theoretical and experimental considerations are needed . we have examined the @xmath1he(in - flight @xmath2,@xmath3 ) reaction at @xmath4 = 1.0 gev / c and @xmath169 for the forthcoming j - parc e15 experiment . we have demonstrated how the spectral shape of the inclusive spectrum is changed by the strength parameters of @xmath7 and @xmath8 in the @xmath2-``@xmath6 '' optical potential , and also obtained the semi - exclusive @xmath2 conversion spectrum which has the ability to be compared with the data by the experiment measuring the missing- and invariant - mass simultaneously . we believe that our considerable assumptions are adequate for the first step toward our purpose . in conclusion , the necessary condition to observe a distinct peak of the @xmath0 bound state turns out to be that @xmath7 is deeper than @xmath156 mev and @xmath8 shallower than @xmath156 mev . the @xmath0 bound state predicted by yamazaki and akaishi @xcite fulfills this condition , while that by shevchenko , gal and mare@xmath17 @xcite stands on the edge of the signal observation in terms of the width . to analyze in detail the forthcoming experimental data at j - parc , it will be necessary to improve our theoretical treatment . this investigation now is in progress @xcite . we acknowledge prof . m. iwasaki for discussions about a plan of the j - parc e15 experiment and for suggestions that a theoretical calculation is needed for its execution . we would like to thank prof . y. akaishi , dr . h. ohta and dr . h. ohnishi for many valuable discussions . this work is supported by grant - in - aid for scientific research on priority areas ( nos . 17070002 and 17070007 ) . 2 for example , a. hirtl , j.marton , e. widmann and j. zmeskal ( eds . ) , proceedings of the _ international conference on exotic atoms and related topics _ , austrian academy of science press , vienna , ( 2005 ) . d. b. kaplan and a. e. nelson , phys , lett . * b175 * ( 1986)57 ; g. e. brown and m. rho , phys . rep . * 269 * ( 1996 ) 333 ; c. h. lee , phys . rep . * 275 * ( 1996 ) 255 . y. nogami , phys . * 7 * ( 1963 ) 288 . t. yamazaki and y. akaishi , phys . b535 * ( 2002 ) 70 ; arxiv:0706.3651 ; proc . academy , series b * 83 * ( 2007 ) 144 . shevchenko , a. gal and j. mare@xmath17 , phys . * 98 * ( 2007 ) 082301 ; n.v . shevchenko , a. gal , j. mare@xmath17 and j. r@xmath157vai , arxiv:0706.4393 . ivanov , p. kleine , j. marton and e. widmann , nucl - th/0512037 . y. ikeda and t. sato , nucl - th/0701001 ; arxiv:0704.1978 . a. dot@xmath157 and w. weise , nucl - th/0701050 . m. agnello @xmath31 @xmath32 , phys . * 94 * ( 2005 ) 212303 . magas , e. oset , a. ramos and h. toki , phys . rev . * c74 * ( 2006 ) 025206 . m. iwasaki , t. nagae @xmath31 @xmath32 , j - parc e15 experiment . the proposal for 1st pac meeting is available from . t. kishimoto , phys . * 83 * ( 1999 ) 4701 ; t. kishimoto @xmath31 @xmath32 , nucl . phys . * a754 * ( 2005 ) 383c . j. yamagata , h. nagahiro and s. hirenzaki , phys . rev . * c74 * ( 2006 ) 014604 . t. yamazaki and y. akaishi , nucl - th/0604049 . k. suzuki @xmath31 @xmath32 , proceedings of the _ international conference on exotic atoms ( exa05 ) _ , austrian academy of science press , vienna , ( 2005 ) 83 . o. morimatsu and k. yazaki , prog . . phys . * 33 * ( 1994 ) 679 . s. tadokoro , h. kobayashi and y. akaishi , phys . rev . * c51 * ( 1995 ) 2656 . t. harada , phys . * 81 * ( 1998 ) 5287 ; nucl . phys . * a672 * ( 2000 ) 181 ; t. harada and y. hirabayashi , nucl . phys . * a759 * ( 2005 ) 143 . j. hfner , s.y . lee and h.a . weidenmller , nucl . * a234 * ( 1974 ) 429 . c.b . dover and a. gal , ann . ( n.y . ) * 146 * ( 1983 ) 309 . i. angeli , atomic data and nuclear data tables * 87 * ( 2004 ) 185 . a. ciepl@xmath170 , e. friedman , a. gal and j. mare@xmath17 , nucl . phys . * a696 * ( 2001 ) 173 . gopal @xmath31 @xmath32 , nucl . b119 * ( 1977 ) 362 . rosenthal and f. tabakin , phys . rev . * c22 * ( 1980 ) 711 . j. mare@xmath17 , e. friedman and a. gal , phys . lett . * b606 * ( 2005 ) 295 ; nucl . a770 * ( 2006 ) 84 . katz , k. bunnel , m. derrik , t. fields , l.g . hyman and g. keys , phys . rev . * d1 * ( 1970 ) 1267 . a. dot@xmath157 , h. horiuchi , y.akaishi and t. yamazaki , phys . rev . * c70 * ( 2004 ) 044313 . t. koike and t. harada , in preparation . t. onaga , h. narumi and t. kouhmura , prog . * 82 * ( 1989 ) 222 . j. yamagata and s. hirenzaki , eur . j. * a31 * ( 2007 ) 255 . he(in - flight @xmath2,@xmath3 ) reaction at @xmath171 gev / c and @xmath5 as a function of the neutron momentum , using the ya optical potential with ( @xmath7 , @xmath8)=(@xmath104 mev , @xmath105 mev ) . the vertical dashed line indicates the corresponding neutron momentum of @xmath172 1224 mev / c at the @xmath2 emitted threshold . the contributions of partial - wave angular momentum states with @xmath119 = 0 , 1 and 2 are also drawn . [ yaspec ] ] , but the inclusive spectrum is decomposed into the conversion processes of @xmath173 ( dashed line ) , and the escape processes of @xmath174 @xmath3 + @xmath2 + `` @xmath6 '' ( dotted line ) . note that the conversion spectrum is identical to the inclusive one in the bound region . [ yaspec2 ] ] he(in - flight @xmath2,@xmath3 ) reaction at @xmath171 gev / c and @xmath5 . the ya optical potential is used . the solid and dashed lines denote the inclusive spectra with and without taking into account the phase space factor , respectively . the dot - dot - dashed line indicates the contribution of the one - nucleon absorption process , @xmath175 , and the dotted line the two - nucleon one , @xmath176 . [ ay - edep ] ] he(in - flight @xmath2,@xmath3 ) reaction at @xmath171 gev / c and @xmath5 on the strength parameter of @xmath7 . @xmath8 is fixed to be @xmath105 mev . the horizontal axis denotes the energy of the @xmath0 system measured from the @xmath2 emitted threshold ; @xmath177 means a bound state of the @xmath0 system . [ svdep ] ]	the formation of a deeply - bound @xmath0 state by the @xmath1he(in - flight @xmath2,@xmath3 ) reaction is investigated theoretically in the distorted - wave impulse approximation using the green s function method . the expected inclusive and semi - exclusive spectra at @xmath4 = 1.0 gev / c and @xmath5 are calculated for the forthcoming j - parc e15 experiment . we employ optical potentials between the @xmath2 and `` @xmath6 '' core - nucleus , and demonstrate systematically the dependence of the spectral shape on @xmath7 and @xmath8 , which are the real and imaginary parts of the strength for the optical potential , respectively . the necessary condition to observe a distinct peak of the @xmath0 bound state with @xmath9 , @xmath10 in the spectrum turns out to be that the value of @xmath7 is deeper than @xmath11 mev and @xmath8 shallower than @xmath11 mev , of which the strength parameters come up to recent theoretical predictions .
in @xcite it was introduced the butterfly presentation of a link diagram as a generalization of the 2-bridge schubert s notation . moreover , the particular concept of a @xmath0-butterfly was implemented in order to study @xmath0-bridge links and to obtain a codification of a @xmath0-bridge link diagram . here , for our purpose , we do not need all the machinery of the butterfly construction presented in @xcite , so we will take a different approach . we will describe the construction of the codification by a direct and combinatorial approach , using the ideas in @xcite , where ferri constructed the crystallization of the double cover of @xmath7 , with a link as ramification set . for any link diagram @xmath8 there is a strong relation between crystallization of the double cover of @xmath7 , with @xmath8 as the ramification set , and the @xmath0-butterfly associated to @xmath8 , that we will explain in @xcite . for any @xmath9-bridge link diagram the construction of an @xmath9-butterfly is possible , see @xcite , but in this paper we want to be specific and we will work only with @xmath0-bridge link diagrams . to a 3-bridge diagram we associate a 3-butterfly that is described by a set of six positive integers @xmath10 , with some restrictions , and then we define the _ schubert form _ of the link diagram as @xmath11 for geometrical reasons that will be explained in section 1 . as each 3-bridge link admits infinitely many different link diagrams , schubert normal form _ for a link @xmath8 is defined by taking the minimum among all 3-bridge link diagrams of @xmath8 according to a lexicographical type of order , see @xcite . for the purpose of this paper we only need the schubert form of the link diagram , but in further research and in the compilation of link tables , it will be interesting to consider the _ schubert normal form _ of a link . in this paper we find formulas for the over and under presentation of the 3-bridge link represented by the schubert form @xmath1 , that depends on the integers @xmath10 . the formula for the under presentation of the 3-bridge link is a natural extension of the formula for the presentation of the 2-bridge link @xmath4 , that depends on the integers @xmath5 and @xmath6 . the paper is organized as follows : in section 1 we describe the construction of a 3-butterfly associated to a 3-bridge link diagram @xmath8 and introduce the schubert form of @xmath8 , that consists of a set of 6 positive integers , @xmath12 , that captures the relevant information of the 3-butterfly and , therefore , of the 3-bridge diagram @xmath8 . in section [ diagram ] we describe a canonical diagram associate to a 3-butterfly @xmath13 , in a similar way to the canonical diagram of a 2-bridge link , see @xcite . in section [ secorien ] we will give an orientation to this canonical diagram . in section [ secpermu ] we define two permutations , @xmath14 and @xmath15 , associated to the schubert form @xmath1 , and study the composition @xmath16 . the cyclic structure of @xmath17 is the key point in the rest of the paper . a variation of the permutation @xmath17 is very useful for the construction of a gauss code for the link diagram , and , from there we can find the dowker code and we are able to compute link invariants , such as the link group , the seifert matrix , the alexander , jones and homfly polynomials . in section [ secgroup ] we present our main result , theorems [ presover ] , [ presoverpal ] and [ presunder ] , that give explicit presentations for the knot group @xmath18 of a @xmath0-bridge link @xmath8 . these presentation are described by clear algorithms , that are easy to program in a computer and depend on the integers in the schubert form of the link diagram . in the last section we propose a special family of links , @xmath19 , that have a strong symmetry that is reflected in the group presentation . this symmetry could be exploited in the study of the representations into sl@xmath20 of the link group . some authors allow that any @xmath9-bridge link diagram can be consider as a @xmath21-bridge link diagram , for any @xmath22 , by considering bridges without any undercrossings , see @xcite . we neither allow this situation nor consider a split link diagram with more than 3 components , as the one in fig . [ fig3]a , as a 3-bridge link diagram we work with 3-bridge link diagrams as in @xcite and @xcite . * remark on notation : * in @xcite the author uses subindexes and denote a butterfly by @xmath23 . in this paper we avoid the use of subindexes in the schubert form , and prefer to assign a different role to each integer , in that way we reach simpler formulas . let @xmath8 be a 3-bridge link such that the projection on the @xmath24 plane is a 3-bridge diagram @xmath25 . let @xmath26 and @xmath27 the bridge projections . draw an ellipse around each of the bridges , in such a way that they are disjoint , and they have the bridges as principal axes . each ellipse will intercept the diagram @xmath25 in an even number of points , that will be the vertices . we denote by @xmath28 and @xmath29 the ellipses around the bridges @xmath26 and @xmath27 , respectively , and let @xmath30(resp . @xmath31 and @xmath32 ) the number of intersections of @xmath33 ( resp . @xmath34 and @xmath35 with the diagram @xmath25 . following @xcite , the ellipses @xmath36 are called butterflies . take the graph @xmath37 formed by the butterflies @xmath36 , the vertices and the bridges . in each butterfly we have the bridge , that divides each butterfly in two halves , that will be the wings . the reflection along the bridges inside the butterflies is called @xmath14 . the segment of the underarcs that are inside the butterflies are forgotten , but they can be recovered with the reflection @xmath14 . the edges outside the butterflies will give the information on how the butterflies intercept to each other , see fig . each one of these edges connect two vertices of two butterflies , we identify these vertices , to form a set that will be the vertices of our graph . the identification will give an involution on the vertex of the graph , that we call @xmath15 . [ ht ] schfigure1.eps we define the @xmath0-_butterfly _ as the graph @xmath38 , formed with the vertices of @xmath39identified by the involution @xmath15 . we draw the graph @xmath40 in any of the three forms shown in fig . if we consider that the diagram @xmath8 is in @xmath41 , then the graph @xmath40 define a polygonalization of @xmath42 formed by three polygons , as shown in fig . compare this simple construction with the formal one given in @xcite and @xcite . when we identify the butterflies , there will appear two new points , that will be denoted @xmath43and@xmath44 . these two points are fundamental , but they are not considered vertices of the @xmath0-butterfly . there can be only two basic forms for the graph @xmath40 , that are determined by the way the butterflies @xmath28 and @xmath29 intersect . [ ht ] schfigure2.eps * type i * : the three butterflies intersect in the two points @xmath43and@xmath44 , see fig . [ fig2].a . * type ii * : two of the polygons do not intersect , see fig . [ fig2].b . when a 3-bridge link diagram produces a type ii butterfly , there is a wave move , see @xcite that allows us to construct a new 3-bridge diagram with lower crossing number . so , we work only with type i butterflies , such that there are no wave moves . in order to obtain a canonical way to describe a @xmath0-butterfly , we will always assume that @xmath45 the condition @xmath46 is to ensure that each bridge has at least one crossing . by rotating the plane and interchanging the points @xmath47 and @xmath48 , we can always obtain a @xmath0-butterfly diagram with @xmath33 at the top , @xmath34 to the left and @xmath29 to the right , and we read it in the counterclockwise direction , @xmath49 , as shown in fig . [ fig1 ] . let @xmath50be the number of vertices between @xmath33 and @xmath34,@xmath51 and @xmath52 . as each butterfly intersects the other two , then @xmath53 and @xmath54 are positive integers that satisfy@xmath55and @xmath56therefore@xmath57 as we will only consider the link diagrams with @xmath58 , then @xmath59 so the integers @xmath60 and @xmath61 satisfy ( [ restr0 ] ) and ( [ restr1 ] ) . reciprocally , if we have integers satisfying ( [ restr0 ] ) and ( [ restr1 ] ) we can construct the butterflies @xmath28 and @xmath29 . now let us describe the positions of the bridges . we orient clockwise each butterfly . from the point @xmath47 , following the orientation , we count the number of vertices between @xmath47 and the vertex in which the bridge begins . we call @xmath62 and @xmath63 the initial points of the bridges in @xmath28 and @xmath29 , respectively . clearly @xmath64 we are working only with link diagrams with exactly three bridges and not only two or one , this impose conditions on the integers @xmath62 and @xmath63 . in @xcite they found the conditions given in the following theorem . [ teoclassi]every @xmath0-butterfly _ defines a unique set of integers _ _ such that _ _ @xmath66 @xmath67 reciprocally , if a set of integers @xmath65 satisfies conditions ( [ resteo ] ) then it defines a @xmath0-butterfly . note that we may think that inside the butterfly @xmath33 ( resp . @xmath68 @xmath29 ) the bridge @xmath69 make a @xmath70 rotation , ( resp . @xmath71 ) . for this geometrical reason we want to use the notation @xmath1 instead of @xmath72 , but @xmath73 is not considered as a rational number . the conditions on the integers @xmath10 impose in ( [ resteo ] ) define a @xmath0-butterfly and a link diagram @xmath8 , but it is possible that @xmath8 is a split link with some trivial components , as the diagram associated to @xmath74 shows , see fig . [ fig3 ] . [ ptb ] schfigure3.eps we say that @xmath13 is a 3-butterfly if the set of integers @xmath10 satisfies the conditions ( [ resteo ] ) . we say that a 3-butterfly is reduced if the associated diagram is a 3-bridge diagram without any trivial components . we say that @xmath75 is a schubert form of a 3-bridge link if the 3-butterfly @xmath13 is reduced . we need to be careful with the relative order of the numbers in @xmath76 . if we change the order , it is possible that we get different schubert forms . the schubert form @xmath77represents a knot and @xmath78 represents a two component link . see fig . we associate to each 3-butterfly @xmath13 a canonical diagram , in a similar way as the canonical diagram of a @xmath3-bridge link is associated to @xmath4 , see @xcite . we draw the three bridges as three segments : bridge @xmath69 as a vertical segment ; bridge @xmath79 as a segment forming a @xmath80 angle with the bridge @xmath69 and bridge @xmath27 as a segment forming a @xmath81 angle with the bridge @xmath69 . we divide the bridge @xmath69 in @xmath5 segments , and we fix two points in each division , one to the left and one to the right , except at the extreme points , where there is only one . label them with @xmath82 , in a counterclockwise sense , so the extreme bridges are labeled @xmath83 and @xmath84 . for the bridge @xmath79 we repeat the process , but we divide the bridge in @xmath6 segments and label the points with @xmath85 . for the bridge @xmath27 the number of segments is @xmath61 and the labels are @xmath86 . the subscripts of @xmath87(resp . @xmath88 and @xmath89 ) are taken @xmath90 , ( resp . @xmath91 and @xmath92 ) . to draw the link diagram we need to join , with appropriate arcs , the points @xmath93and @xmath94 , @xmath95 and @xmath96 according to the rules given by permutations @xmath15 and @xmath14 . there are @xmath97 arcs between the @xmath69 and @xmath79 bridges , namely @xmath98 likewise there are @xmath99 arcs between the @xmath79 and @xmath27 bridges , that are@xmath100 and , finally , @xmath101 arcs between the @xmath27 and @xmath69 bridges,@xmath102 it is enough to know how to construct the first arc between each pair of bridges , and the rest of the arcs are `` _ _ parallel _ _ '' arcs to them , see fig . [ fig4 ] [ ht ] schfigure4.eps in the rest of this paper we will refer to the diagram described as the link canonical diagram associated to the schubert form @xmath103 . notice that if @xmath1 does not satisfy the conditions in theorem [ teoclassi ] , we still may use this algorithm to draw a link diagram . if the schubert form @xmath1 is reduced , the @xmath0-bridge diagram has @xmath104 crossings . the conditions for a 3-butterfly @xmath10 to be reduced can not be described using simple conditions on the integers in a similar way as the conditions to be a 3-butterfly given in ( [ resteo ] ) . now we need to go deeper and study in detail the permutations @xmath15 and @xmath14 . given a set @xmath10 that satisfies ( [ resteo ] ) we construct explicitly the associated @xmath0-butterfly and then we draw the 3-bridge diagram . define the 3-butterfly by labelling the vertices of each of the butterflies : @xmath33 have vertices labeled by @xmath105 @xmath34 with vertices @xmath106 and @xmath29 with vertices @xmath107 @xmath108 the bridge ends are labeled by @xmath83 and @xmath84 in @xmath33 ( resp . by @xmath109 and @xmath110 in @xmath34 and @xmath111 and @xmath112 in @xmath29 ) . [ fig5 ] . [ ptb ] schfigure5.eps we have the permutations @xmath14 and @xmath113on the set @xmath114 . the permutation @xmath14 is the reflection along the bridges . the permutation @xmath15 is determined by the identification of the vertices of two butterflies , so in the @xmath0-butterfly each vertex has two labels . the proofs of the following lemmas are straightforward computations . [ lemmagama]the function defined in the set @xmath114 by @xmath115{ccc}\gamma\left ( a_{i}\right ) = a_{2p - i } , & & 0\leq i<2p,\ \ \\ \gamma\left ( b_{j}\right ) = b_{2q - j } , & & 0\leq j<2q,\ \ \\ \gamma\left ( c_{h}\right ) = b_{2s - h } , & & 0\leq h<2s\ \ \end{array } \label{permugama}\ ] ] is an order 2 permutation . the set of fixed points is@xmath116 the set @xmath117 corresponds to the endpoints of the bridges . it will play an important role in the rest of the paper . [ lemmafi]the map @xmath118 @xmath119 defined by @xmath115{llll}a_{n - i } & \longleftrightarrow & b_{m+i-1 , } & \text { if } 1\leq i\leq t,\\ a_{n+j } & \longleftrightarrow & c_{l - j-1 , } & \text { if } 0\leq j\leq w-1,\\ b_{m - h } & \longleftrightarrow & c_{l+h-1 , } & \text { if } 1\leq h\leq v , \end{array } \label{permufi}\ ] ] is an order 2 permutation , where @xmath120 @xmath99 and @xmath101 . note that @xmath15 does not have fixed points , and among the bicycles in @xmath15 there is no a bicycle in the set @xmath121 the construction of @xmath15 is well defined for any polygonalization of @xmath42 with @xmath0 polygons , even if they do not satisfy the conditions of theorem [ teoclassi ] . in fact , in terms of the permutation @xmath15 , we can rewrite theorem [ teoclassi ] as follows . a set @xmath10 , with @xmath122describes a 3-butterfly if and only if the associated permutation @xmath15 does not have any of the bicycles in the set @xmath123 . we study the cyclic decomposition of @xmath124 . the orbit of a vertex @xmath125 will be denoted by @xmath126 . for a cycle @xmath127 we will use the same symbol to refer to the cycle , to its orbit @xmath128 and to the word @xmath129 . the length of @xmath130 will be denoted by @xmath131 will denote the cycle that contains @xmath132 and for a function @xmath133 @xmath134 will be the word ( set ) formed by applying @xmath135 to each element in @xmath130 . [ teored]let @xmath13 be a 3-butterfly , @xmath14 and @xmath15 be its associated permutations , given in lemmas [ lemmagama ] and [ lemmafi ] and let @xmath124 . @xmath1 is a schubert form for a 3-bridge link if and only if @xmath17 is the product of three disjoint cycles,@xmath136 such that , for i=1,2,3 , @xmath137 , where @xmath138 is given in ( [ extrem ] ) . let @xmath124 associated to the 3-butterfly @xmath12 . the 3-butterfly@xmath139 defines a schubert form @xmath1 if and only if the 3-butterfly is reduced . suppose that the butterfly is reduced . the orbit of @xmath140under @xmath141 will describe a path that follows the underarc with initial point in @xmath84 , so eventually it will arrive to the endpoint of the underarc , say @xmath142 , @xmath143 @xmath144 @xmath145defined in ( [ extrem ] ) . then @xmath146 , so the orbit will go back to the same underarc , in opposite direction . we take @xmath147 as the cycle formed by the orbit of @xmath84 and @xmath148 will contain exactly two vertices . we repeat the same process with the other vertices in @xmath138 . since the butterfly is reduced , all the vertices will be crossed by one of the underarcs , so we have only three orbits . reciprocally , if the butterfly is not reduced , there will be a component whose vertex will not be in the orbit of any of the elements in @xmath138 . [ fig3]a . from now on we will assume that the permutation @xmath17 associated to the schubert form @xmath1 is the product of three disjoint cycles , @xmath149 . the cyclic decomposition of @xmath17 allows us to determine the number of components of the associated link diagram . [ classification][teoclaslink]let @xmath1 be a schubert form , @xmath14 and @xmath15 its associated permutations given in [ lemmagama ] and [ lemmafi ] , @xmath124 . the @xmath0-bridge link diagram @xmath8 represented by @xmath1 satisfies : \(i ) @xmath8 is a knot if and only if @xmath150 and @xmath151 . \(ii ) @xmath8 is a two component link if and only if one , and only one , of the following conditions holds : @xmath152 or @xmath153 . \(iii ) @xmath8 is a three component link if and only if @xmath154 and @xmath153 . take the cyclic decomposition of@xmath136 and study each one of the cycles , using the interpretation given in the proof of theorem [ teored ] . until now we have not considered the orientation of the link @xmath8 , but in order to find a group presentation for the link group @xmath18 we will give an orientation to the canonical diagram described in section [ diagram ] . let @xmath155 , we study in detail these cycles . in each cycle @xmath156 we have two special vertices , that are the fixed points of @xmath14 and form the set @xmath138 defined in ( [ extrem ] ) . each cycle describes a path around one of the link diagram underarcs , see fig . [ fig6]._b _ , so one of this special vertices corresponds to the arc initial point , denoted @xmath157 ; and the other one to the arc endpoint , denoted @xmath158 . so we consider that when we follow the link , we travel it in the order @xmath159 and @xmath160 and the bridge @xmath69 in the direction from @xmath83 to @xmath84 . we define @xmath161 ( resp . @xmath162 ) , the direction in which we travel the bridge@xmath163 ( resp . @xmath164 ) as : @xmath165 and @xmath115{ccc}\delta_{b}=\left\ { \begin{array } [ c]{cc}1 , & \text{if we go from } b_{0}\ \text{to } b_{q}\text{\ } \\ -1 , & \text{if we go from } b_{q}\ \text{to } b_{0}\end{array } \right . , & \ & \delta_{c}=\left\ { \begin{array } [ c]{cc}1 , & \text{if we go from } c_{0}\ \text{to } c_{s}\text{\ } \\ -1 , & \text{if we go from } c_{s}\ \text{to } c_{0}\end{array } \right . \end{array}\ ] ] when the condition ( i ) in theorem [ teoclaslink ] is satisfied , the schubert form corresponds to a knot diagram , hence the orientation of bridge @xmath69 is enough to determine the knot orientation . we take @xmath166as the cycle that contains @xmath84 and @xmath160 as the cycle that contains @xmath83 . in the link case we need to determine the orientation of each component . if @xmath8 is a 3-component link , the condition ( iii ) in theorem [ teoclaslink ] is satisfied and we orient each component by @xmath167 , @xmath147 contains @xmath168 contains @xmath110 and @xmath169 contains @xmath112 . if @xmath8 is a 2-component link the condition ( ii ) in theorem [ teoclaslink ] holds , again we take @xmath166as the cycle that contains @xmath84 , and @xmath160 the cycle that corresponds to the other component . if @xmath170is a knot , table 1 contains all possibilities for the endpoints of the cycles @xmath171 and the knot orientation . \b . if @xmath170is a 2-component link , table 2 contains all possibilities for the endpoints of the cycles @xmath171 and the link orientation . @xmath115{cc}\begin{tabular } [ c]{\|l\|l\|l\|l\|l\|l\|l\|l\|}\hline $ i_{1}$ & $ f_{1}$ & $ i_{2}$ & $ f_{2}$ & $ i_{3}$ & $ f_{3}$ & $ \delta_{b}$ & $ \delta_{c}$\\\hline\hline $ a_{p}$ & $ b_{0}$ & $ b_{q}$ & $ c_{0}$ & $ c_{s}$ & $ a_{0}$ & $ 1 $ & $ 1$\\\hline $ a_{p}$ & $ b_{0}$ & $ b_{q}$ & $ c_{s}$ & $ c_{0}$ & $ a_{0}$ & $ 1 $ & $ -1$\\\hline $ a_{p}$ & $ b_{q}$ & $ b_{0}$ & $ c_{0}$ & $ c_{s}$ & $ a_{0}$ & $ -1 $ & $ 1$\\\hline $ a_{p}$ & $ b_{q}$ & $ b_{0}$ & $ c_{s}$ & $ c_{0}$ & $ a_{0}$ & $ -1 $ & $ -1$\\\hline $ a_{p}$ & $ c_{0}$ & $ c_{s}$ & $ b_{0}$ & $ b_{q}$ & $ a_{0}$ & $ 1 $ & $ 1$\\\hline $ a_{p}$ & $ c_{0}$ & $ c_{s}$ & $ b_{q}$ & $ b_{0}$ & $ a_{0}$ & $ -1 $ & $ 1$\\\hline $ a_{p}$ & $ c_{s}$ & $ c_{0}$ & $ b_{0}$ & $ b_{q}$ & $ a_{0}$ & $ 1 $ & $ -1$\\\hline $ a_{p}$ & $ c_{s}$ & $ c_{0}$ & $ b_{q}$ & $ b_{0}$ & $ a_{0}$ & $ -1 $ & $ -1$\\\hline \end{tabular } & \begin{tabular } [ c]{\|l\|l\|l\|l\|l\|l\|l\|l\|}\hline $ i_{1}$ & $ f_{1}$ & $ i_{2}$ & $ f_{2}$ & $ i_{3}$ & $ f_{3}$ & $ \delta_{b}$ & $ \delta_{c}$\\\hline\hline $ a_{p}$ & $ b_{0}$ & $ b_{q}$ & $ a_{0}$ & $ c_{s}$ & $ c_{0}$ & $ 1 $ & $ 1$\\\hline $ a_{p}$ & $ b_{q}$ & $ b_{0}$ & $ a_{0}$ & $ c_{s}$ & $ c_{0}$ & $ -1 $ & $ 1$\\\hline $ a_{p}$ & $ c_{0}$ & $ c_{s}$ & $ a_{0}$ & $ b_{q}$ & $ b_{0}$ & $ 1 $ & $ 1$\\\hline $ a_{p}$ & $ c_{s}$ & $ c_{0}$ & $ a_{0}$ & $ b_{q}$ & $ b_{0}$ & $ 1 $ & $ -1$\\\hline $ a_{p}$ & $ a_{0}$ & $ b_{q}$ & $ c_{0}$ & $ c_{s}$ & $ b_{0}$ & $ 1 $ & $ 1$\\\hline $ a_{p}$ & $ a_{0}$ & $ b_{q}$ & $ c_{s}$ & $ c_{0}$ & $ b_{0}$ & $ 1 $ & $ -1$\\\hline \end{tabular } \\ \text { table 1\ } & \text{table 2}\end{array}\ ] ] to avoid the lack of uniqueness in the cycles , we always write the cycle @xmath156 as an ordered set with initial point @xmath157 , but to simplify notation we keep the cycle notation . in general this will not generate confusion in our work . [ lemcycle]let @xmath17 be the permutation associated to the schubert form @xmath76 , @xmath149 , for i=1,2,3 we have : \(i ) each cycle @xmath156 is even , with order @xmath172greater than 4 . \(ii ) @xmath173for @xmath174 \(iii ) @xmath175contains the transposition @xmath176 . by condition ( [ forbiden ] ) we get @xmath177 , so the length of the cycle @xmath156 is greater than 3 . then , there exists @xmath178 @xmath179 such that @xmath180 and @xmath181 this yields@xmath182 now,@xmath183 and then , for @xmath184 we get @xmath185 the relevant information on each cycle @xmath156 is contained in the first part of the cycle , we define the initial segment of @xmath156 as@xmath186 we may summarize the results up to now in an algorithm that allows us to find the cycles @xmath171 , the set @xmath138 and therefore the directions @xmath187 and @xmath188 associated to a schubert form . let @xmath189 be the permutation in @xmath114 defined by @xmath190 this permutation corresponds to _ `` travel the bridges '' _ in the diagram . [ algorithm]given a schubert form @xmath1 the following algorithm finds the orientation of the associated link diagram . it provides the cycles @xmath171 such that @xmath191 where the cycle @xmath156 has the form given in lemma [ lemcycle ] . take @xmath192 and @xmath193 \2 . if @xmath194 take @xmath195 else take @xmath196 \3 . take @xmath197 and @xmath198 \4 . if @xmath199 then take @xmath200 else take @xmath201 as the unique element in @xmath202 . take @xmath203 and @xmath204 . let @xmath8 be the link diagram with schubert form @xmath1 . we have an explicit way to find the over and under presentations of the link group of @xmath8 , see @xcite and @xcite . this method requires to use the link diagram . we will use this method , but we will replace the explicit use of the diagram by an algorithm that uses the permutations @xmath205 and @xmath17 and some new functions defined on the set @xmath114 . as the description of the over and under presentations requires an oriented link diagram , we will always refer to the standard link diagram and orientation described in section [ diagram ] . it is important to remark that we need the diagram only to explain the construction , but the algorithm to find the presentations do not require to draw the link diagram , it depends only on the permutations @xmath205 and @xmath124 . as @xmath205 depend only of the schubert form @xmath1 , the presentation of the link group will depend only on the integers @xmath10 . the algorithm is efficient and easy to implement in a software such as _ mathematica . _ we take meridians around the bridges as group generators , and label them by the same name as the bridges , so we have generators @xmath26 and @xmath27 , see fig . [ fig6]_a . _ [ h ] schfigure6.eps we find the relators by traveling the frontier of a neighborhood of the underarcs , as shown in fig . [ fig6]_b _ , so these paths are precisely the orbits @xmath206 . each relator is a word in @xmath26 and @xmath27 constructed with the convention that each time we cross the bridge @xmath69 ( resp . @xmath164 ) we write @xmath207 ( resp . @xmath208 ) depending of the sign of the crossing , given by the convention schfigure7.eps . we replace this graphic process by defining a function @xmath135 that `` _ _ forgets the index but remembers the direction _ _ '' . consider @xmath209 defined by@xmath210{cc}a & \text{if } 0<i\leq p\\ a^{-1 } & \text{otherwise}\end{array } \right . , \gamma\left ( b_{i}\right ) = \left\ { \begin{array } [ c]{cc}b^{\delta_{b } } & \text{if } 0<i\leq q\\ b^{-\delta_{b } } & \text{otherwise.}\end{array } \right . , \label{gamrel}\\ \gamma\left ( c_{i}\right ) & = \left\ { \begin{array } [ c]{cc}c^{\delta_{c } } & \text{if } 0<i\leq s\\ c^{-\delta_{c } } & \text{otherwise.}\end{array } \right . \nonumber\end{aligned}\ ] ] the relators are @xmath211 and @xmath212 , where @xmath213 means the word obtained when we apply @xmath135 to each element in the orbit @xmath156 . thus we have proved the following proposition . [ presover]the link group of the link @xmath8 given by the schubert form @xmath1 admits a presentation given by @xmath214 were @xmath149 is the associated permutation and @xmath135 is given in ( [ gamrel ] ) . by the symmetry of the cycles described in lemma [ lemcycle ] , we may rewrite the relators as the relations . when the schubert form @xmath75 defines a knot , using the information in table 1 we find that @xmath215 or@xmath216 for the case when it is a link we have similar relations . at this moment we have lost the geometrical meaning of the generators , so we may rename the generators , if necessary , and unify the two cases , so we have the following proposition . [ presoverpal]the link @xmath8 given by the schubert form @xmath75 admits a presentation given by @xmath217 were @xmath218 is a word in @xmath219 given by@xmath220 were @xmath135 is defined in ( [ gamrel ] ) and @xmath221 is defined in ( [ defonda ] ) . we know that in the case of knots , one of the relations is redundant , but for practical reasons we prefer to have all of them and , in particular computations , we omit the longest relation , that we do not know in advance which one will be.@xmath222this contrasts with the under presentation that we will introduce in the next section , in which we know the lengths of the words involved . the sum of the lengths of the words @xmath223 and @xmath224 is @xmath225 when @xmath8 is a knot , the peripheral system of the group is given by @xmath226 with @xmath227 and @xmath21 is the exponent sum of the word @xmath228 . by direct computation we have @xmath229 the under presentation is the dual presentation of the over presentation . those dual presentations play a central role in the proofs of properties of the knot group and the alexander polynomial of knots , see @xcite . by studying these presentations we find a similar algorithm to the known one to find the 2-bridge link group , that has an explicit formula depending of @xmath5 and @xmath6 . of course , we need a more elaborate algorithm . we take as generators of @xmath18 the meridians around the underarcs , see [ fig5]_c . _ again , the key point is to use the cyclic decomposition of @xmath17 . we call @xmath69 ( resp . @xmath79 and @xmath230 the generators corresponding to the underarc described by @xmath147 ( resp . @xmath231 and @xmath160 ) . the relations are given by traveling the boundary of each butterfly , that describe simple closed paths around the bridges . so the first path is given by @xmath232 , the second by @xmath233 and the third by @xmath234 . the graphical procedure to find the relators is : each time we cross the link we encounter one of the vertices in the set @xmath235 we identify the underarc , say @xmath132 , and the sign of the crossing , @xmath236 and write @xmath237 , with @xmath238 and @xmath239 . again , this procedure will be established by defining a function@xmath240 , similar to the one defined in ( [ gamrel ] ) , that identifies the underarc that contains the vertex and the direction of the crossing . let @xmath241 defined by @xmath242{ll}a & \text{if } x\in\widetilde{\tau_{1}}\\ a^{-1 } & \text{if } x\notin\widetilde{\tau_{1}}\ \\ b^{\delta_{b } } & \text{if } x\in\widetilde{\tau_{2}}\\ b^{-\delta_{b } } & \text{if } x\notin\widetilde{\tau_{2}}\ \\ c^{\delta_{c } } & \text{if } x\in\widetilde{\tau_{3}}\\ c^{-\delta_{s } } & \text{if } x\notin\widetilde{\tau_{3}}\ \end{array } \right . \ if\ } x\notin e,\ \rho\left ( x\right ) = \left\ { \begin{array } [ c]{ll}a & \text{if } x\in\widetilde{\tau_{1}}\\ a^{-1 } & \text{if } \gamma\left ( x\right ) \in\widetilde{\tau_{1}}\ \\ b^{\delta_{b } } & \text{if } x\in\widetilde{\tau_{2}}\\ b^{-\delta_{b } } & \text{if } \gamma\left ( x\right ) \in\widetilde{\tau_{2}}\ \\ c^{\delta_{c } } & \text{if } x\in\widetilde{\tau_{3}}\\ c^{-\delta_{s } } & \text{if } \gamma\left ( x\right ) \in\widetilde{\tau_{3}}\ \end{array } \right.\ ] ] [ defro ] the relators are @xmath243 for the symmetry of the functions and the cycles given in lemma [ lemcycle ] , we have that if @xmath244 @xmath245 , therefore if we take the words@xmath246 the relators become the relations @xmath247 or@xmath248 note that the lengths of the words @xmath249 and @xmath250 are @xmath251 @xmath252 and @xmath253,@xmath222respectively . in this case it is not possible to change the variable names , because we want to have the information about the word lengths . now the peripheral system is given by @xmath254 where @xmath255 , were @xmath256 is the exponent sum of the word @xmath257 . we have proved the following theorem : [ presunder]the link @xmath8 given by the butterfly @xmath75 admits a presentation given by : \i . if @xmath8 is a knot @xmath258 \ii . if @xmath8 is a 3-component link @xmath259 with @xmath249 and @xmath250 words of length @xmath251 @xmath252 and @xmath253,@xmath222respectively , defined by ( [ defpalu ] ) . if @xmath8 is a 2-component link there are six possible combination for the presentation , that are the natural variations of @xmath260 * note : * this construction does not depend on the fact that @xmath261 , nor that we are working with type i butterfly . so we may use it in a more general way . however , if we take the schubert form @xmath75 we know that in the knot case one of the relations is redundant , and in this presentation we know that the longest is the first one , so usually that is the one we eliminate . in general we do not have an exact pattern for a 3-bridge link group , as the one we encounter for 2-bridge links , see @xcite , but there are families of 3-bridge links with a very regular pattern for the fundamental group . one of them is the family of links with schubert form @xmath262 , for integers @xmath263 . this family contains : borromean rings @xmath264 , the pretzel link @xmath265 , that corresponds to @xmath266 ; the toroidal knot @xmath267 and its mirror image @xmath268 , that corresponds to @xmath269 and to @xmath270 , respectively . the standard diagrams of the links in this family have symmetries of order 2 and 3 . the knot @xmath280 in rolfsen 's table has schubert normal form@xmath281 and @xmath282 . note that it is the toroidal knot @xmath283 . in general , the toroidal link @xmath284 is a 3-component link if @xmath285 and it is a knot in the other cases ; and the word @xmath54 in proposition [ proppalw ] is @xmath286{cc}\left ( zyx\right ) ^{p/3 } & \text{if } p\equiv1\ \operatorname{mod}3\\ \left ( zyx\right ) ^{[p/3]}z & \text{if } p\equiv2\ \operatorname{mod}3\\ \left ( zyx\right ) ^{[p/3]}zy & \text{if } p\equiv0\ \operatorname{mod}3 , \end{array } \right.\ ] ] were @xmath287 $ ] means the integer part of @xmath288 . the knot @xmath289 in rolfsen 's table has schubert normal form@xmath290 and @xmath291 . note that it is the pretzel knot @xmath292 . in general the pretzel link @xmath293 is a knot if @xmath5 is odd and a 3-component link if @xmath5 is even and the word @xmath54 in proposition [ proppalw ] is @xmath286{cc}\left ( z^{-1}x\right ) ^{\left [ p/2\right ] } z^{-1}\left ( z^{-1}y\right ) ^{\left [ p/2\right ] } & \text{if } p\ \text{is odd}\\ \left ( z^{-1}x\right ) ^{(p-2)/2}z^{-1}xy^{-1}\left ( xy^{-1}\right ) ^{\left ( p-2\right ) /2 } & \text{if } p\ \text{is even.}\end{array } \right.\ ] ]	we introduce the schubert form a @xmath0-bridge link diagram , as a generalization of the schubert normal form of a @xmath0-bridge link . it consists of a set of six positive integers , written as @xmath1 , with some conditions and it is based on the concept of @xmath0-butterfly . using the schubert normal form of a @xmath0-bridge link diagram , we give two presentations of the 3-bridge link group . these presentations are given by concrete formulas that depend on the integers @xmath2 the construction is a generalization of the form the link group presentation of the @xmath3-bridge link @xmath4 depends on the integers @xmath5 and @xmath6 .
metal - organic systems with a 3d@xmath6 electron configuration , such as in cu@xmath7 , are expected to undergo a jahn - teller distortion . in octahedral coordination , this typically elongates one axis of the octahedron and removes the degeneracy of the e@xmath8 orbitals d@xmath9 and d@xmath10 . this effect widely occurs in molecular systems as well as in other materials such as the colossal magnetoresistance manganites.@xcite the jahn - teller distortion is extremely sensitive to bond distances and as such applied pressure can strongly influence or even induce a jahn - teller distortion . depending on the elongation axis in the crystal structure , the pattern of orbital overlaps and related exchange interactions can vary significantly in turn leading to disparate magnetic ground states . finding materials where a jahn - teller distortion can be tuned to act as a magnetic switch has potential for applications in technological devices and , thus , is of great interest . the copper - based coordination polymer magnet cuf@xmath0(h@xmath0o)@xmath0(pyz ) ( pyz = pyrazine ) appears to be a model material for studying the switching of magnetic properties due to changes in the jahn - teller axis.@xcite cuf@xmath0(h@xmath0o)@xmath0(pyz ) crystallizes in a monoclinic structure ( space group p2@xmath11/c ) with a=7.6926 , b=7.5568 , c=6.897 and @xmath12=111.065@xmath13 under ambient conditions.@xcite the structure consists of cuf@xmath0o@xmath0n@xmath0 octahedra which form a 2d network in the bc plane . the magnetic susceptibility shows a broad peak at 10.5 k consistent with two dimensional ( 2d ) short range correlations and three dimensional long - range order below t@xmath14 = 2.6 k due to weak coupling along the a - axis . at ambient pressure , the jahn - teller axis is along the n - cu - n axis ( a direction ) . hence the d@xmath9 magnetic orbital lies in the bc plane to form a two dimensional ( 2d ) antiferromagnetic ( afm ) quasi square lattice where the near neighbor distances are the same , but the interior angles deviate by @xmath1 @xmath55@xmath15 . first principles electronic structure calculations confirm this configuration.@xcite these calculations give an intra - plane afm exchange interaction of about 13 to 19 k and an inter - plane afm exchange interaction of only 1% of the intra - plane exchange interaction . an important feature of cuf@xmath0(h@xmath0o)@xmath0(pyz ) is that under an applied pressure of 0.9 gpa , a switch of the jahn - teller axis from the n - cu - n to the o - cu - o ( c direction ) bond occurs.@xcite when pressure increases to 3.1 gpa , the jahn - teller axis switches again from o - cu - o bond to f - cu - f bond ( b direction ) . correspondingly the magnetic interactions are expected to vary at different pressures : from the ambient pressure 2d quasi square lattice ( exchange path cu - f@xmath16h - o - cu ) to one dimensional ( 1d ) chain interactions ( exchange path cu - pyz - cu ) at 0.9 gpa . indeed , the magnetic susceptibility at ambient pressure can be well described by the 2 d heisenberg square lattice @xcite while above 0.9 gpa the magnetic susceptibility exhibits a broad peak consistent with 1d magnetic correlations@xcite . therefore a microscopic understanding of the magnetic properties at ambient pressure would be an illuminating step towards a complete understanding of the interesting pressure dependent behavior of cuf@xmath0(h@xmath0o)@xmath0(pyz ) . in this paper we present the results of a study of deuterated cuf@xmath0(h@xmath0o)@xmath0(pyz ) ( cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) ) with elastic and inelastic neutron scattering under ambient pressure . the neutron diffraction results show that the sample has a collinear afm structure with the moments lying in ac - plane along the [ 0.7 0 1 ] direction . the spin wave dispersion extracted from the inelastic neutron spectra can be well described by a nearest - neighbor 2d heisenberg model . the spin wave dispersion found along the inter - layer direction could not be observed within the instrumental resolution of fwhm=0.034 mev . this indicates that the inter - layer exchange interaction @xmath4 is weak as expected for a 2d system where the spin wave dispersion does not depend on the out of plane direction . consequently , cuf@xmath0(h@xmath0o)@xmath0(pyz ) is a good example of quasi-2d system where the spin wave dispersion depends exclusively on k and l. within the 2d magnetic plane , the dispersion along the zone boundary points has been measured and is in accord with prediction of the quantum monte carlo and series expansion calculations@xcite for quantum fluctuation corrections to linear spin wave theory for the s=1/2 square lattice heisenberg antiferromagnet . in order to minimize attenuation of the beam caused by the large incoherent scattering cross - section of hydrogen , fully deuterated cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) crystals were synthesized for this study . substitution of deuterium for hydrogen does not alter the crystal symmetry and induces only small changes in the lattice parameters.@xcite to ensure that structural parameters of the samples studied here are in accord with previously published values@xcite neutron diffraction data ( not shown ) were collected on the four circle neutron diffractometer hb-3a at the high flux isotope reactor ( hfir ) , oak ridge national laboratory . the magnetic structure was studied with the thermal triple - axis spectrometer hb-1 at the hfir . a crystal with a mass of 0.01 g was studied using hb-3a while a larger crystal with mass 0.1 g was studied with hb-1 . a silicon monochromator and no analyzer were used for the measurements on hb-3a and unless otherwise noted all triple - axis measurements were performed using pyrolytic graphite ( 002 ) monochromator and analyzer crystals . inelastic neutron scattering experiments were performed using the cold neutron triple - axis spectrometer cg-4c ( hfir ) , the thermal triple - axis hb-1a ( hfir ) and the cold neutron chopper spectrometer cncs@xcite at the spallation neutron source ( sns ) , oak ridge national laboratory . the inelastic neutron scattering experiments were performed on four co - aligned crystals with total mass of 0.85 g and total mosaic of 0.7@xmath13 . for the cg-4c experiment , measurements were made with fixed final energy , @xmath18 , of 5 mev and 3.7 mev resulting in an elastic energy resolution of about 0.3 mev and 0.17 mev , respectively . for the hb-1a experiment , the incident energy @xmath19 was fixed at 14.7 mev with a pyrolytic graphite ( 002 ) monochromator and a beryllium ( 002 ) analyzer was used to obtain an elastic energy resolution of about 0.58 mev . for the cncs measurements , the incident energy @xmath19 was fixed at 3 mev and 1.5 mev to allow examination of the entire spin wave spectrum . the energy resolution ( fwhm ) was about 0.08 mev and 0.034 mev at the elastic position for @xmath19 = 3 mev and 1.5 mev , respectively . ( a ) temperature dependence of magnetic bragg peaks ( a ) : ( 0.5 1 0 ) and ( b ) : ( -0.5 0 1 ) for cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) . insets display the temperature dependent peak intensity . solid lines are guides to the eye . the data were collected on the hb-1 triple - axis instrument in @xmath20@xmath210 and @xmath200@xmath22 scattering planes , respectively.,scaledwidth=40.0% ] temperature dependent neutron diffraction reveal the presence of additional peaks for temperatures less than 2.5 k at wave vectors q=(0.5 @xmath21 @xmath22 ) with integer values of @xmath21 and @xmath22 and where @xmath21+@xmath22 is an odd number . this indicates that in real space the magnetic lattice is doubled along the a direction compared to nuclear lattice while in the bc - plane the nuclear lattice and magnetic lattice remain the same size . 1 shows the temperature dependent magnetic bragg peaks at both ( 0.5 1 0 ) and ( -0.5 0 1 ) collected on the hb-1 instrument using the final energy @xmath18=13.5 mev(@xmath23=2.4617 ) . in the insets the temperature dependent peak intensity for ( 0.5 1 0 ) and ( -0.5 0 1 ) are presented . a power - law fit of the form ( 1-t / t@xmath14)@xmath24 gives a @xmath25 of 2.51 @xmath1 0.06 k and a value of @xmath12 of 0.24 @xmath1 0.07 . this low value of @xmath12 is likely consistent with 2d critical behavior , however , care should be used when interpreting this finding as a more rigorous series of measurements is required to definitively establish the critical exponents . the value of t@xmath14 determined from the powerlaw fit is slightly smaller than the value reported by muon spin relaxation measurements of a non - deuterated sample@xcite but is in reasonable agreement with the value of @xmath25= 2.53 k determined from magnetic susceptibility measurements of fully deuterated samples.@xcite the observed intensity plotted as a function of calculated intensity for cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) at t=0.25 k. nuclear and magnetic reflections are denoted by diamonds and circles , respectively . the solid line is a guide to the eye . the data were collected on hb-1 in different scattering planes . the insets depict different views of the unit cell . the arrows on cu atoms indicates the magnetic moment configuration . the 2d magnetic behavior of cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) originates in the bc plane . , scaledwidth=47.0% ] a total of 26 nuclear and 30 magnetic reflections have been collected in @xmath260 and @xmath200@xmath22 scattering planes at t=0.3 k using hb-1 . representation analysis for the general case of propagation vector ( 0.5 0 0 ) in the structure symmetry @xmath27 have been performed using sarah - representational analysis.@xcite the results show that there are two possible irreducible representations as shown in table i. using the basis vectors in table i , assuming the magnetic moment direction @xmath28=(@xmath29 @xmath30 @xmath31 ) varies freely , the magnetic reflections can be fitted . for both cases , when @xmath30 @xmath32 0 , the spins will configure in noncollinear way . fitting of moment direction was carried out by minimizing @xmath33 ^ 2\ ] ] + where n is the number of reflections . fits considering both representations were performed and the irreducible representation denoted by @xmath34 yielded the best agreement with the data . the resulting collinear spin structure has moments along the [ 0.7 0 1 ] real space direction . for simplicity , the dipole approximation to the cu@xmath7 form factor has been used . this may lead to a systematic error in both the direction and magnitude of the magnetic moment due to the anisotropy of the d@xmath9 orbital.@xcite we note that the @xmath34 irreducible representation with m@xmath35=0 results in an antiparallel alignment of the spins on cu site 1 ( atomic coordinate ( 0 0 0 ) ) and cu site 2 ( atomic coordinate ( 0 0.5 0.5 ) ) consistent with the observed propagation vector which requires k+l = odd . the resulting magnetic structure for cuf@xmath0(h@xmath0o)@xmath0(pyz ) , shown in fig . 2 , is g - type ( nearest neighbor ) antiferromagnetic order . .[tab : table ] basis vectors for cu site 1 ( atomic coordinates ( 0 0 0 ) ) and cu site 2 ( atomic coordinates ( 0 0.5 0.5 ) ) determined from the representational analysis@xcite for space group no . 14 ( p2@xmath11/c ) and magnetic propagation vector ( 0.5 0 0 ) . here ir represents irreducible representations , bv represents basis vectors . [ cols="^,^,^,^",options="header " , ] interestingly the moment direction [ 0.7 0 1 ] in real space does not appear to be along an obvious structural direction . the selection of a specific moment direction suggests some anisotropy in the spin hamiltonian . the absolute value of the magnetic moment m was determined by comparing the intensity of magnetic and nuclear bragg reflections . to carefully estimate the moment size , the instrument resolution was taken into account using reslib@xcite and the debye - waller factor has been included . our calculations yield an ordered moment of 0.60 @xmath1 0.03 @xmath2/cu . this value is much smaller than the spin-@xmath36 free ion moment . in low dimensional systems a moment reduction occurs due to quantum fluctuations and has been observed in a number of materials.@xcite the observed and calculated intensities of magnetic and nuclear reflections together with the resulting spin arrangement are presented in fig . ( a ) constant-@xmath37 scans along ( 0.5 @xmath21 0 ) collected in the @xmath260 scattering plane at t = 1.5 k by the cg-4c spectrometer . ( b ) constant-@xmath38 scans collected in the @xmath260 scattering plane at t = 1.5 k by the cg-4c spectrometer . the solid lines in both panels are fits with gaussians . , scaledwidth=50.0% ] the spin wave dispersion has been extracted from inelastic neutron scattering measurements . we first discuss inelastic neutron scattering data collected using the triple - axis spectrometer cg-4c in the @xmath20@xmath210 scattering plane . both constant-@xmath38 and constant-@xmath37 scans in the @xmath260 scattering plane have been performed . constant-@xmath37 scans at several different energy transfer are plotted in fig . 3(a ) and constant-@xmath38 scans are shown in fig . these measurements were carried out along the ( 0.5 @xmath21 0 ) direction where k=1 corresponds to the magnetic zone center . the excitation energy at the zone boundary point @xmath38=(0.5 1.5 0 ) is @[email protected] mev . within instrumental resolution ( [email protected] mev ) , no energy gap at the antiferromagnetic zone center has been observed . constant - q scans along the zone boundary points ranging from ( 0 1 -0.5 ) to ( 0 0.5 -0.5 ) . the solid line are fits to a gaussian . the data were collected on hb-1a . ] in figs . 4(a ) , ( b ) and ( c ) the spin wave dynamics have been studied with the sample aligned in the @xmath200@xmath22 scattering plane with the time - of - flight spectrometer cncs . 4(a ) shows that the spin wave excitation is essentially independent of @xmath20 at the excitation energy @xmath40=0.25 mev , implying only weak interactions along @xmath20 . in fig . 4(b ) , the full dispersion along the @xmath22-direction is shown . the zone boundary energy in this direction is @[email protected] mev . this value is consistent with the previously determined zone boundary energy along the @xmath21-direction ( @xmath39 ) . to search for a small energy gap , measurements were performed on cncs using an incident energy of 1.5 mev resulting in an energy resolution of [email protected] mev ) . the measurements were centered on the ( 0.5 0 1 ) magnetic zone center . the data obtained along the @xmath22-direction is shown in fig . again , within instrumental resolution ( fwhm=0.034 mev ) , no energy gap in @xmath22 dispersion curve was observed . furthermore , no dispersion was observed along the @xmath20-direction for l=1 . we note that previous dft calculations predict a inter - layer exchange constant of 0.011 mev @xcite and the resulting interlayer dispersion could not be observed with our instrumental resolution . nearly identical exchange interactions along the @xmath21 and @xmath22 directions together with the lack of dispersion along the @xmath20-direction indicates the magnetism in cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) is predominately 2d . the inter - layer @xmath20-direction magnetic exchange was also probed with time - of - flight spectrometer cncs in a high resolution mode with @xmath19 = 1.5 mev . the @xmath20 vs. @xmath37 slice , for l=1 ( not shown here ) is featureless and no obvious spin excitation has been observed within instrument resolution . due to the fact that in spin wave theory the zone boundary energy of the spin wave excitation is of order 2@xmath42 , our data show that the inter - layer exchange interaction @xmath4 is expected to be smaller than @xmath43 mev = 0.017 mev . considering the @xmath41=2.204 mev which gives the intra - layer @xmath42 of 1.102 mev , it is reasonable to estimate that the inter - layer exchange interaction is less than 1.5% of the intra - layer exchange interaction . this ratio is in agreement with that determined by both dft calculations and by fitting a heisenberg square lattice model to the magnetic susceptibility which in both cases yields @xmath4/@xmath3 @xmath5 1%.@xcite since no energy gap has been observed in the 2d square lattice plane at the magnetic zone center within instrumental resolution it is reasonable to believe that any spin exchange anisotropy must be very small . however , the magnetic moments select the [ 0.7 0 1 ] direction suggesting nevertheless that a spin anisotropy is present . as cu@xmath7 moments should have no single ion anisotropy and the presence of inversion symmetry prohibits dzyaloshinskii - moriya interactions , we have included exchange anisotropy of the following form : @xmath44\ ] ] where the summation is over the nearest neighbors in the 2d plane , j is the effective intra - layer exchange parameter and @xmath45 is the exchange anisotropy parameter . linear spin - wave theory yields the spin wave dispersion @xmath46 where @xmath38 is a function of k and l. the dashed lines in the inset of fig . 6 indicate the diamond pattern on which the cu - ions are positioned . under the approximation that the diamond is replaced by a square , the abscissa labels in fig . 6 are related to square lattice notation in the following way : ( 0.5 1 0 ) and ( 0.5 0 1 ) correspond to ( @xmath47 ) ; ( 0.5 1.5 0 ) and ( 0.5 0 1.5 ) to ( @xmath48/2 @xmath48/2 ) ) ; and finally the point ( 0 0.5 -0.5 ) to ( @xmath48 0 ) . in fig . 6 we plot the extracted dispersion relation for cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) . the experimental data in the 2d plane were determined from combinations of triple - axis and time - of - flight measurements . the solid line in fig . 6 is the expected dispersion behavior of the classical heisenberg model ( @xmath45 = 1 ) with @xmath42= 1.102 @xmath1 0.003 mev . in the s=@xmath36 heisenberg square lattice antiferromagnet , it is well established that the spin dynamics can be well described by the classical linear spin wave theory with the inclusion of quantum corrections.@xcite the net effect of quantum corrections is an overall renormalization factor , @xmath49 1.18 , resulting in an effective coupling constant @xmath50 = @xmath51 when considering the dispersion from the ( @xmath48 @xmath48 ) zone center to the ( @xmath48/2 @xmath48/2 ) zone boundary . thus , our extracted exchange constant , @xmath42 , is an effective coupling constant and the resulting @xmath3 is 0.934 @xmath1 0.0025 mev within the square lattice approximation . this value is consistent with estimates from the magnetic susceptibility and magnetization measurements of fully deuterated samples where @xmath3 is about 0.94 mev.@xcite the dispersion resulting from the anisotropic exchange yields a zone center energy gap of @xmath52 = @xmath53 . as noted previously , we have not observed such a gap within the instrumental resolution of 0.034 mev . this allows us to place a lower bound on @xmath45 . the coupling constant extracted from the zone boundary measurements is @xmath42= 1.102 @xmath1 0.003 mev . placing an upper bound on the gap energy of 0.034 mev results in a value of @xmath45 ( the ratio @xmath54 ) of at least 0.99988 . the result would be a very small anisotropy resulting in a slightly larger @xmath55 which would favor spin orientation along the @xmath56-axis . note that the @xmath56-axis is the spin wave quantization direction , in this case the real - space [ 0.7 0 1 ] direction . this same spin anisotropy may be related to the observed low - field spin - flop like transition observed for cuf@xmath0(h@xmath0o)@xmath0(pyz ) @xcite the spin wave dispersion has been examined along the antiferromagnetic zone boundary . in fig . 5 we present constant - q scans along ( 0 @xmath21 -0.5 ) from @xmath21=1 ( square lattice ( @xmath48/2 @xmath48/2 ) ) to @xmath21=0.5 ( square lattice ( @xmath48 0 ) ) . in fig . 6 the observed dispersion between ( 0 1 -0.5 ) and ( 0 0.5 -0.5 ) has been plotted . compared to ( 0 1 -0.5 ) , the excitation energy at ( 0 0.5 -0.5 ) has been suppressed by [email protected]% . this observation is inconsistent with linear spin wave theory where there is no dispersion along the zone boundary . however , proper inclusion of quantum fluctuations using series expansion and quantum monte carlo techniques for the s=1/2 2d afm square lattice predict a dispersion along the zone boundary with the energy at ( @xmath48 0 ) 7 - 10% lower than that at ( @xmath572 @xmath48/2).@xcite experimental behavior consistent with these calculations have been seen in several model systems.@xcite however , there are some materials where conflicting experimental observations exist . for instance , in la@xmath0cuo@xmath17@xcite the energy scale at ( @xmath48 0 ) is about 7% larger than the value at ( @xmath48/2 @xmath48/2 ) in contrast to the expected quantum behavior which has been attributed to ring exchange among four spins in the cuo@xmath0 plane . in the coordination polymer compound cu(pyz)@xmath0(clo@xmath17)@xmath0,@xcite the excitation energy at ( @xmath48 0 ) is reduced by 11.5% compared to the energy at ( @xmath48/2 @xmath48/2 ) . the authors believe this value is larger than the expectation of series expansion and quantum monte carlo calculations and they attribute this stronger suppression to the next - nearest - neighbor interactions which enhances quantum fluctuations . in k@xmath0v@xmath58o@xmath59@xcite a striking feature of two modes near the zone boundary point ( @xmath48/2 @xmath48/2 ) has been observed experimentally and no fully satisfactory explanation has yet been given for this behavior . for our measurements on cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) , in fig . 6 the dash - dotted line is the results of calculations including quantum fluctuations with nearest neighbor interactions@xcite . it is clear that the data points follow the expected theory and , therefore , we interpret the zone boundary dispersion behavior in cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) as being consistent with the s=1/2 near neighbor 2d afm square lattice with inclusion of quantum corrections . summary of the spin wave dispersion for cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) . the data were collected using cncs ( solid circles ) , cg-4c ( solid squares ) and hb-1a ( solid triangles ) . the solid line represents the dispersion of heisenberg linear spin wave theory with a nearest neighbor interaction . the dash - dotted line represents the results of series expansion to higher order.@xcite the inset depicts the bc - plane . the dashed lines show the square lattice with a nearest neighbor exchange interaction . , scaledwidth=45.0% ] we have performed a series of elastic and inelastic neutron scattering experiments to study the 2d spin-@xmath36 quasi square lattice antiferromagnet cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) . the 2d magnetic lattice in the bc - plane is the same size as nuclear structure lattice while along the @xmath60 direction the magnetic lattice is doubled . the ordered magnetic moment of the cu@xmath7 ions is found to be 0.60 @xmath1 0.03 @xmath2 which is significantly reduced from the expected 1 @xmath2 , indicating strong quantum fluctuations . the spins adopt a collinear antiferromagnetic alignment for nearest neighbor sites in the 2d plane and are oriented along the real - space [ 0.7 0 1 ] direction . the spin dynamics of cuf@xmath0(d@xmath0o)@xmath0(d@xmath17-pyz ) have been studied using inelastic neutron scattering data obtained from both triple - axis and time - of - flight instruments . the 2d spin wave dispersion can be described by a nearly isotropic spin hamiltonian with small nearest neighbor interaction @xmath3 = 0.934 @xmath1 0.0025 mev and very weak inter - layer coupling . along the magnetic zone boundary , the excitation energy shows a [email protected]% dispersion consistent with square lattice calculations including quantum fluctuations . research work at ornl was sponsored by the laboratory directed research and development program of ornl , and was supported by the scientific user facilities division , office of basic energy sciences , doe . this research was sponsored by the division of materials science and engineering of the u.s . department of energy ( rsf ) . this work was supported by uchicago argonne , llc , operator of argonne national laboratory ( `` argonne '' ) . argonne , a u.s . department of energy office of science laboratory , is operated under contract no . de - ac02 - 06ch11357 . work at ewu was supported by the nsf under grant no . dmr-1005825 . j. l. manson , m. m. conner , j. a. schlueter , a. c. mcconnell , h. i. southerland , i. malfant , t. lancaster , s. j. blundell , m. l. brooks , f. l. pratt , j. singleton , r. d. mcdonald , c. lee , m .- h . whangbo , chem . mater . * 20 * , 7408 ( 2008 ) . p. a. goddard , j. singleton , c. maitland , s. j. blundell , t. lancaster , p. j. baker , r. d. mcdonald , s. cox , p. sengupta , j. l. manson , k. a. funk , and j. a. schlueter , phys . b * 78 * , 052408 ( 2008 ) . y. j. kim , a. aharony , r. j. birgeneau , f. c. chou , o. entin - wohlman , r. w. erwin , m. greven , a. b. harris , m. a. kastner , i. ya . korenblit , y. s. lee , and g. shirane , phys . lett . * 83 * , 852 ( 1999 ) . n. tsyrulin , t. pardini , r. r. p. singh , f. xiao , p. link , a. schneidewind , a. hiess , c. p. landee , m. m. turnbull , and m. kenzelmann , phys . lett . * 102 * , 197201 ( 2009 ) ; n. tsyrulin , f. xiao , a. schneidewind , p. link , h. m. rnnow , j. gavilano , c. p. landee , m. m. turnbull , and m. kenzelmann , phys . b * 81 * , 134409 ( 2010 ) .	we have performed elastic and inelastic neutron scattering experiments on single crystal samples of the coordination polymer compound cuf@xmath0(h@xmath0o)@xmath0(pyz ) ( pyz = pyrazine ) to study the magnetic structure and excitations . the elastic neutron diffraction measurements indicate a collinear antiferromagnetic structure with moments oriented along the [ 0.7 0 1 ] real - space direction and an ordered moment of 0.60 @xmath1 0.03 @xmath2/cu . this value is significantly smaller than the single ion magnetic moment , reflecting the presence of strong quantum fluctuations . the spin wave dispersion from magnetic zone center to the zone boundary points ( 0.5 1.5 0 ) and ( 0.5 0 1.5 ) can be described by a two dimensional heisenberg model with a nearest neighbor magnetic exchange constant @xmath3 = 0.934 @xmath1 0.0025 mev . the inter - layer interaction @xmath4 in this compound is less than 1.5% of @xmath3 . the spin excitation energy at the ( 0.5 0.5 0.5 ) zone boundary point is reduced when compared to the ( 0.5 1 0.5 ) zone boundary point by @xmath5 10.3 @xmath1 1.4 % . this zone boundary dispersion is consistent with quantum monte carlo and series expansion calculations for the s=1/2 heisenberg square lattice antiferromagnet which include corrections for quantum fluctuations to linear spin wave theory .
analytic solutions for the edge mode branches on a semi - infinite plane can be obtained by a straightforward calculation . we consider a semi - infinite plane where @xmath186 is a superfluid region and @xmath187 is vacuum . our hamiltonian for @xmath186 reads @xmath188.\end{aligned}\ ] ] analytic properties of a similar hamiltonian with @xmath19 have been discussed in the context of the topological insulators @xcite . we first discuss the @xmath6-wave sf and then move to the @xmath15-wave sf . the edge mode wavefunction @xmath189 with a boundary condition @xmath190 is given by @xmath191(e^{\lambda_1y}-e^{\lambda_2y}),\\ \lambda_{1,2}^2&=k_x^2 + 2m_0\left ( \frac{2m_0\delta^2}{2}-\mu\right)\notag \\ & \quad \pm2m_0\sqrt{\left(\frac{2m_0\delta^2}{2}-\mu\right)^2 -\mu^2+e^2 } , \label{eq : lam}\end{aligned}\ ] ] where re@xmath192 so that @xmath193 . since @xmath194 , non - zero @xmath195 lead to two equations @xmath196 and @xmath197 , where @xmath198 by solving these equations , we obtain @xmath199 from eqs . ( [ eq : lam ] ) and ( [ eq : lam _ ] ) , the edge mode energy is given as @xmath200 ^ 2=\delta_0 ^ 2k_x^2/k_f^2.\end{aligned}\ ] ] note that this is valid only for re@xmath192 , or equivalently @xmath201 . from @xmath202 and @xmath203 together with the normalization of @xmath189 , the components of @xmath189 are determined as @xmath204 where @xmath205 . for the @xmath15-wave sf , we can compute the edge mode branches in the same way . in this case , @xmath202 and @xmath203 are determined by @xmath206 ^ 2+\delta^2}+k_x^2\notag\\ & \qquad \pm\frac{1}{\sqrt{[1/(2m_0)]^2+\delta^2 } } \sqrt{\frac{-\mu^2\delta^2}{[1/(2m_0)]^2+\delta^2}+e^2},\\ & \lambda_{1}+\lambda_2=\frac{2k_xe}{2k_x^2/(2m_0)-\mu},\\ & \lambda_{1}\lambda_2=2m_0\mu+\frac{2m_0e^2}{2k_x^2/(2m_0)-\mu}-k_x^2.\end{aligned}\ ] ] from these equations , we obtain @xmath207 ^ 2=\frac{\delta_0 ^ 2\varepsilon_f^2}{\varepsilon_f^2+\delta_0 ^ 2 } \left(2\frac{k_x^2}{k_f^2}-\frac{\mu}{\varepsilon_f}\right)^2 . \label{eq : ed}\end{aligned}\ ] ] the above @xmath203 is reduced to the well - known quasi - classical result @xmath208 when @xmath209 @xcite . the components @xmath210 are the solutions of @xmath211 it is easy to check @xmath212 and @xmath213 which are implied from the particle - hole symmetry of the hamiltonian , @xmath214 with @xmath215 . note that @xmath216 is restricted to @xmath217 from the condition re@xmath192 and eq . ( [ eq : a2b2 ] ) . in the quasi - classical limit , @xmath218 are simply reduced to @xmath219 . here , we discuss @xmath63 in the quasi - classi limit @xmath220 in the bcs regimes . first , as we discussed in the main text and saw in fig . 2 ( b ) , @xmath67 changes @xmath2212 only when a non - phs edge mode branch crosses zero energy as a function of @xmath46 , which is confirmed numerically . therefore , we simply count how many times the non - phs edge mode branches cut zero energy , and obtain @xmath222}\theta(l - l_j)\end{aligned}\ ] ] for @xmath223 , where @xmath147 is the `` fermi angular momenta '' at which @xmath224-th edge mode branch crosses zero energy as a function of @xmath46 and @xmath225 is the step function . we have labeled the edge modes so that @xmath226 is satisfied . @xmath227 holds for @xmath228 by the particle - hole symmetry . by using this , we obtain @xmath229 } 2\times \frac{1}{2}(l_j+\nu/2)(l_j+\nu/2 - 1)\notag \\ & \simeq -\frac{1}{2}\sum_{j=1}^{\nu}l_j^2 \label{eq : ll}\end{aligned}\ ] ] for even @xmath4s . in the second equality , we neglected @xmath230 contributions which is much smaller than the @xmath231-contributions when the total number of fermions @xmath2 is large enough . contributions from @xmath228 has been included by introducing a factor 2 in front of the summation @xmath232}$ ] . equation ( [ eq : ll ] ) also holds for odd @xmath4s . the fermi angular momentum @xmath233 of the edge modes are evaluated within quasi - classical calculations . when the radius of the system is large , @xmath234 , curvature of the boundary could be negligible and the physics around the boundary would be equivalent to that in a cylinder system or a semi - infinite system . in such systems , edge mode wavefunctions are proportional to @xmath235 $ ] where @xmath236 are a wavenumber and a position along the boundary , respectively . since the edge modes are well localized around the boundary , we can regard the edge modes as running at the boundary @xmath237 in the disc geometry . in this case , the ( anti-)periodic boundary condition is satisfied , @xmath238 . especially , the fermi angular momentum @xmath239 of the edge modes are connected with their fermi wavenumbers @xmath240 as @xmath241 therefore , we obtain a quasi - classical expression for @xmath63 , @xmath242 the fermi wavenumbers @xmath240 can easily be calculated for semi - infinite systems within the quasi - classical calculations . let us assume that the system has a straight boundary along @xmath243-axis at @xmath244 which separates the sf in @xmath245 from the vacuum in @xmath246 . by solving the bdg equation near the boundary , we obtain an equation which determines the edge mode energy @xmath203 @xcite , @xmath247 here , we have defined @xmath248 with @xmath249 . if we rewrite @xmath250 where @xmath251 are odd / even with respect to @xmath252 , the above equation leads to @xmath253 at the fermi wavenumber of the one dimensional edge modes where @xmath254 . by using @xmath255 , it is seen that @xmath256 for an even @xmath4 , and @xmath257 for an odd @xmath4 . then , we can evaluate the fermi wavenumbers of the @xmath4 edge modes along the @xmath243-direction . it is written as @xmath258 where @xmath259 for @xmath260 satisfying @xmath261 . for @xmath7 , since @xmath262 we obtain an identity within the quasi - classical calculation , @xmath263 ^ 2\notag\\ & = k_f^2\sum_{j=1}^{\nu}\left(\frac{1}{2}-\frac{\cos(2\theta_{f}^{(j ) } ) } { 2}\right)\notag\\ & = \nu \frac{k_f^2}{2}. \label{eq : quasi_id}\end{aligned}\ ] ] there would be possible deviations of @xmath148 from the above values , which are beyond the present quasi - classical approximation . from eq . , we obtain the same identity for the perpendicular component @xmath264 , @xmath265 these equations are a direct consequence of the residual @xmath4-hold rotational symmetry of the gap function . the condition @xmath253 means vanishing edge mode energy @xmath254 when the two - dimensional wavenumber @xmath266 is at nodes of @xmath267 , where positions of the nodes are @xmath4-hold rotationally symmetric . this symmetry immediately leads to @xmath268 , from which we can easily reproduce eqs . and since the wavenumber is on the two - dimensional fermi surface , @xmath269=k_f^2 $ ] for each @xmath224 within the quasi - classical calculation . on the other hand , for @xmath19 , there is no residual rotational symmetry in @xmath267 . this explains the difference between @xmath19 and @xmath7 . we can see that eqs . and are indeed satisfied by directly looking at @xmath148 ; @xmath270 for @xmath20 , @xmath271 for @xmath133 , and @xmath272 for @xmath136 , and so on . we have numcerically confirmed that eq . holds also for the disc geometry . for example , when @xmath273 , the edge mode fermi wavenumbers are numerically obtained as @xmath274 for @xmath20 , latexmath:[$\|k_{f\parallel}^{(1,3)}\|\simeq68k_f/80\simeq 0.85k_f , @xmath136 for several values of @xmath277 . we note that , although the above discussions are not based on self - consistent calculations , eq . should be satisfied even in self - consistent calculations for the bcs limit , since @xmath240 are generally determined by symmetry of the gap functions . when @xmath33 and @xmath75 are tuned so that @xmath2 is fixed , the total number of fermions on the disc is expressed as @xmath278 where @xmath279 is the particle density in two dimensions and @xmath280 is the volume of the system . we have neglected @xmath281 contributions which can arise from deviations of the particle density from @xmath282 near the boundary . from eqs . ( [ eq : quasi_l ] ) , ( [ eq : quasi_id ] ) , and ( [ eq : bcs_n ] ) , we obtain eq . ( 6 ) in the main text . a general theory of the bogoliubov transformation was developed in ref . we follow it and obtain the ground state wavefunctions for the chiral sfs on the disc . the bogoliubov transformation is introduced as @xmath283 = \sum_{m=1}^{m}\left [ \begin{array}{cc } u_{nm}^{(l ) } & u_{nm+m}^{(l)}\\ v_{nm}^{(l ) } & v_{nm+m}^{(l)}\\ \end{array } \right ] \left [ \begin{array}{c } b_{m}^{(l)}\\ b_{m+m}^{(l)}\\ \end{array } \right],\end{aligned}\ ] ] where @xmath284 are eigenvectors of the bdg equation , @xmath285= e_{m}^{(l ) } \left [ \begin{array}{c } u_{nm}^{(l)}\\ v_{nm}^{(l)}\\ \end{array } \right].\end{aligned}\ ] ] we have introduced the cut - off @xmath286 as in the main text . the ground state @xmath62 is defined as a vacuum for all the quasi - particle excitations with positive energies , @xmath287 let the number of the positive ( negative ) eigenvalues of @xmath71 be @xmath288 which are @xmath289 in general , and we label the eigenvalues so that @xmath290 . correspondingly , the @xmath291-operators are rewritten as @xmath292 for @xmath293 and @xmath294 for @xmath295 . we also rewrite the unitary matrix for the transformation as , @xmath296^{-1 } = \left [ \begin{array}{cc } u^{(l)}_1 & u^{(l)}_2\\ u^{(l)}_3 & u^{(l)}_4\\ \end{array } \right],\end{aligned}\ ] ] where sizes of the matrices are @xmath297 for @xmath298 and @xmath299 for @xmath300 . since @xmath301 is an @xmath302 hermitian matrix , it has real eigenvalues @xmath303 with eigenvectors @xmath304 , @xmath305 we denote the number of @xmath306 as @xmath307 . @xmath308 is a @xmath309 hermitian matrix and has real eigenvalues @xmath310 with eigenvectors @xmath311 . similarly , an @xmath302 hermitian matrix @xmath312 has real eigenvalues @xmath313 with eigenvectors @xmath314 , and a @xmath315 hermitian matrix @xmath316 has real eigenvalues @xmath317 with eigenvectors @xmath318 . we denote the number of @xmath319 as @xmath320 . we then define new fermionic operators , @xmath321 the above transformations can be implemented by a unitary operator @xmath322 which transforms the @xmath323 operators as @xmath324 an explicit form of the unitary operator @xmath322 is found in ref . @xcite . by definition , @xmath325 , i.e. @xmath326 , are satisfied in the transformation . the ground state wavefunction is now obtained as @xmath327 , where @xmath328 the coefficients @xmath329 are the solution of @xmath330 for @xmath331 . it is noted that at least one of @xmath332 or @xmath333 must be zero in our hamiltonian , because the present gap functions are `` nodeless '' . in order to see this , we first divide the hamiltonian as @xmath334 , where @xmath335 \tilde{h}_{{\rm bdg}jj'}^{(l ) } \left [ \begin{array}{c } \tilde{c}_{j'l+\nu\uparrow}\\ \tilde{c}^{\dagger}_{j'-l\downarrow}\\ \end{array } \right],\\ \tilde{h}_{{\rm bdg}}^{(l)}&= \left [ \begin{array}{cc } x_3^{(l)\ast}&0\\ 0&x_2^{(l)}\\ \end{array } \right]^{\dagger } h_{{\rm bdg}}^{(l ) } \left [ \begin{array}{cc } x_3^{(l)\ast}&0\\ 0&x_2^{(l)}\\ \end{array } \right]\notag\\ & = \left [ \begin{array}{cc } \tilde{\varepsilon}^{(l)}&\tilde{\delta}^{(l)}\\ \tilde{\delta}^{(l)\dagger}&-\tilde{\varepsilon}'^{(l)}\\ \end{array } \right].\end{aligned}\ ] ] each of @xmath336 ( @xmath337 ) consists of some combinations of @xmath338 , and @xmath339 . @xmath340 includes the @xmath323-operators only for @xmath341 , while @xmath342 has the @xmath323-operators only for @xmath343 . @xmath344 connects the two sectors . since @xmath345 is partly diagonal in the @xmath224-space , the off - diagonal part vanishes @xmath346 . therefore , latexmath:[$h_{1,2}\|{\rm gs}\rangle_l \propto here , let us assume that both of @xmath168 and @xmath169 are non - zero . then , from @xmath348 for @xmath349 and @xmath350 . similarly , from @xmath351 for @xmath352 and @xmath350 , and @xmath353 for @xmath354 and @xmath355 . this means that @xmath356 is not invertible , and @xmath357 is also non - invertible . therefore , @xmath358 must have an eigenvector with zero - eigenvalue . since this matrix is a representation of the hermitian operator @xmath359 on @xmath360\times [ 0,2\pi))$ ] with the dirichlet boundary condition @xmath361 , @xmath359 also must have a zero - eigenvale . however , since laplacian is positive definite on @xmath360\times [ 0,2\pi))$ ] , @xmath358 does not have an eigenvector with zero - eigenvalue , which means that @xmath362 and @xmath363 are invertible . because the resulting non - invertibility of @xmath356 under the above assumption contradicts with this , at least one of @xmath332 or @xmath333 must be zero . this is reasonable since it can be considered that the factor @xmath364 corresponds to nodes of gap functions , which is seen in the ground state for nodal sfs under the periodic boundary condition , latexmath:[$\|{\rm nodal}\rangle \propto \prod_{k:\delta_k=0,k < k_f}c^{\dagger}_{k\uparrow } c^{\dagger}_{-k\downarrow } \prod_{k:\delta_k\neq 0}\exp \left(-v_k / u_kc^{\dagger}_{k\uparrow}c^{\dagger}_{-k\downarrow}\right ) we can evaluate @xmath178 within the present scheme , either by using the relation @xmath366 or by directly using the explicit expression of @xmath62 . then , it is confirmed that eq . ( 3 ) and eq . ( 8) in the main text are indeed equivalent . this equivalence is used in order to determine @xmath367 in the following . in the @xmath6-wave sf , @xmath102 for all @xmath46 both in the bec regime and the bcs regime . therefore , the ground state wavefunction has only the exponential factor and is given by , for any @xmath33 , @xmath368\|0\rangle.\end{aligned}\ ] ] for simplicity , we have suppressed the indices @xmath369 . similarly , in the bec regimes with @xmath7 , since @xmath102 holds , @xmath370 and the ground state wavefunctions are simply of the exponetial forms . however , in the bcs regimes with @xmath7 , there exist the factor @xmath371 or @xmath372 . since @xmath373 and @xmath374 , we can determine each of them from the relation @xmath375 . for the @xmath15-wave states , since @xmath112 for @xmath113 , we have @xmath376 and @xmath377 there . similary , since @xmath114 for @xmath115 , we have @xmath378 and @xmath379 . hence , the ground state wavefunction is given by @xmath380 \end{array } \right . \label{eq : dgs}\end{aligned}\ ] ] we note that , @xmath381-fermions have lower energies than @xmath382-fermions with the same radial quantum number @xmath383 for @xmath384 . this would be the reason why the @xmath385-fermions are unpaired for @xmath384 , but not vice versa . similar wavefunctions with unpaired fermions represented by extra @xmath386 operators are also obtained for @xmath387 in the bcs regimes .	we study the orbital angular momentum ( oam ) @xmath0 in two dimensional chiral @xmath1-wave superfluids ( sf ) of @xmath2 fermions on a disc at zero temperature , in terms of spectral asymmetry and spectral flow . it is shown that @xmath3 for any integer @xmath4 , in the bec regime . in contrast , in the bcs limit , while the oam is @xmath5 for the @xmath6-wave sf , for chiral sf with @xmath7 , the oam is remarkably suppressed as @xmath8 , where @xmath9 is the gap amplitude and @xmath10 is the fermi energy . we demonstrate that the difference between the @xmath6-wave sf and the other chiral sfs in the bcs regimes originates from the nature of edge modes and related depairing effects . the orbital angular momentum ( oam ) @xmath11 of chiral superfluids ( sf ) of fermions is a fundamental problem which has been under intense investigation over several decades @xcite . among the chiral sfs , the simplest @xmath6-wave pairing state describes the a - phase of liquid @xmath12he @xcite , and quite likely also the superconducting phase of sr@xmath13ruo@xmath14 @xcite . the oam is a direct manifestation of the broken chiral symmetry @xcite , and is also closely related to the edge current , which has been so far not observed experimentally in sr@xmath13ruo@xmath14 despite the expectation from the @xmath6-wave sf picture @xcite . most of the existing studies on the oam have focused on @xmath6-wave sf . however , higher - order chiral sfs such as @xmath15 , @xmath16 , -wave ones are also of interest @xcite , and have potential applications to candidates for chiral superconductors , such as upt@xmath17 @xcite , uru@xmath13si@xmath13 @xcite , and srptas @xcite . we focus on fundamental chiral sfs with the pairing symmetry @xmath18 which can be classified by the integer angular momentum @xmath4 of each cooper pair : @xmath19 corresponds to @xmath6 , @xmath20 to @xmath15 , and so on . in fact , as we will demonstrate in this letter , there is an unexpected fundamental difference between the @xmath6-wave and higher - order chiral sfs with respect to the oam ; thus it is essential to consider the higher - order ones as well , for a complete understanding of the problem . the main issue with the oam is that different viewpoints lead to different predictions for it , resulting in an apparent paradox @xcite . one argument is that , since each cooper pair has the oam @xmath4 , @xmath21 where @xmath2 is the total number of fermions . there is , however , a different view starting from the normal ( non - superconducting ) fermi liquid , which has @xmath22 . formation of cooper pairs with an angular momentum would change the value of @xmath11 from zero to a non - vanishing value in the chiral sf phase . however , since only the low - energy fermions near the original fermi surface would be affected , @xmath11 should be suppressed as @xmath23 with @xmath24 , where @xmath9 is the pairing gap amplitude and @xmath10 is the fermi energy . of course , the analysis did not stop at the hand - waving arguments and many calculations have been carried out based on various schemes , leading to different results . we note that , in the limit of strong pairing of fermions , the superfluid phase may be understood as a result of bose - einstein condensation ( bec ) of bosonic molecules . in this limit , it would be natural to expect that @xmath21 , since each bosonic molecule carries the oam @xmath4 . however , this does not necessarily imply that the same value of @xmath11 persists in the regime where the superfluid is described by bardeen - cooper - schrieffer ( bcs ) theory . in fact , the `` weak - pairing '' chiral sfs in the bcs regime is a topological superfluid with gapless edge states , while the `` strong - pairing '' chiral sfs in the bec regime is a non - topological one @xcite . thus they are distinct superfluid phases separated by a quantum phase transition , and thus @xmath11 could take very different values . even within the same phase , the stability of the oam has not been established . historically , strong reduction of @xmath11 ( @xmath25 ) was predicted for the bcs regime of @xmath6-wave sf , in several of the earlier papers on the subject @xcite . on the other hand , many others , including most of more recent ones support the full oam @xmath26 at zero temperature , even for the bcs regime @xcite . however , so far there is no clear physical picture why @xmath26 holds even in the bcs regime . experimental investigation of the problem is difficult and there have been very few reports so far @xcite . therefore the long - standing paradox is not yet resolved even for the @xmath6-wave sf , let alone for the higher - order ones with @xmath27 . in this study , we investigate the problem in the simplest ideal setting : two dimensional ( 2d ) chiral sfs confined on a completely circular disc with a specular wall at zero temperature , in the framework of bogoliubov - de gennes ( bdg ) hamiltonian . for simplicity , we assume that the @xmath28-vector is @xmath29 for the triplet states so that both the singlet states and the triplet states can be discussed in a parallel way ; our analysis is also applicable to the spinless fermions with slight modifications . we consider the hamiltonian @xmath30 \psi_{\sigma } + \int d^2x\psi_{\uparrow}^{\dagger}\delta(r)(p_x+ip_y)^{\nu } \psi_{\downarrow}^{\dagger}+({\rm h.c.}),$ ] where @xmath31 , @xmath32 is the fermion mass , and @xmath33 is the chemical potential . @xmath34 describes the wall of the container , and is chosen to be @xmath35 and @xmath36 with a radius @xmath37 . there is no texture in this model . although we only consider constant pair potentials in numerical calculations , our discussion is also applicable to systems with @xmath38 for which @xmath39-dependence is determined through self - consistent calculations . the field operator is expanded in terms of a single particle basis as @xmath40 where @xmath41 satisfies @xmath42\varphi_{nl}=\varepsilon_{nl } \varphi_{nl}$ ] @xcite . then the hamiltonian becomes @xmath43^t\notag \\ & \quad \times \left [ \begin{array}{cc } \varepsilon_{n , l+\nu}\delta_{nn^{\prime } } & \delta_{nn^{\prime}}^{(l)}\\ \delta_{n^{\prime}n}^{(l)\ast } & -\varepsilon_{n ,- l}\delta_{nn^{\prime } } \end{array}\right ] \left [ \begin{array}{c } c_{n^{\prime},l+\nu,\uparrow}\\ c_{n^{\prime},-l,\downarrow}^{\dagger } \end{array}\right ] , \label{eq : ham}\end{aligned}\ ] ] where @xmath44 , with an appropriate high - energy regularization . we denote the above matrix as @xmath45 . the particle - hole symmetry connects different @xmath46-sectors as @xmath47 where @xmath48 in the nambu space for odd ( even ) @xmath4 , which implies that , although eigenvalues come in pairs , each of them lies in different @xmath46-sectors . the oam @xmath11 corresponds to the operator @xmath49 , while the total particle number operator is given as @xmath50 . these operators are clearly defined for the present model , and include all the possible contributions . neither @xmath51 nor @xmath52 commutes with the bdg hamiltonian owing to the pairing term in the hamiltonian , and is not conserved . nevertheless , as pointed out in refs . @xcite , the combination @xmath53 commutes with the hamiltonian , and thus is a conserved quantity . physically , @xmath54 represents the correction to the oam with respect to its `` full '' value @xmath55 . if the ground state belongs to the zero eigenvalue sector of @xmath56 , it follows that @xmath57 . however , @xmath54 could take different eigenvalues in the ground state , as it is clear by considering the limit of @xmath58 , where @xmath59 and @xmath60 hold . thus the eigenvalue @xmath61 of @xmath54 in the ground state is a nontrivial quantity . in fact , it can still be calculated exactly for the hamiltonian . after the bogoliubov transformation , the ground state @xmath62 is simply the vacuum with respect to all the positive energy quasiparticles . the eigenvalue @xmath63 for @xmath62 can be obtained explicitly as @xmath64 where @xmath65 are eigenvalues of @xmath66 , and @xmath67 is called the spectral asymmetry @xcite . in actual calculations , we introduce a cut - off @xmath68 so that indices @xmath69 are restricted not to exceed @xmath70 . then @xmath71 is represented by a @xmath72 matrix , and @xmath67 takes only even integer values . we have verified that the results are independent of @xmath70 , when @xmath70 is sufficiently large . from this formula , it is clear that @xmath63 can change only when there is a spectral flow , namely some of the eigenvalues of @xmath71 cross zero as model parameters are varied . we first discuss the @xmath6-wave states ( @xmath19 ) for which the spectrum is particle - hole symmetric about @xmath73 : @xmath74 for simplicity , we treat the two parameters @xmath33 and @xmath75 independently as in refs . @xcite for a discussion of the spectral flow . in the bec regime , by numerically diagonalizing @xmath71 , we obtain fully gapped spectrum as shown in fig . [ fig : p+ip](a ) , and find that the spectral asymmetry @xmath67 is zero for all @xmath46 . as a consequence , @xmath76 and thus follows @xmath77 , as expected . however , the spectrum becomes less trivial if the system is in the bcs regime . there , a single edge mode with @xmath78 appears as seen in fig . [ fig : p+ip](c ) , refecting the topological nature of the phase @xcite . this edge mode is particle - hole symmetric by itself , @xmath79 , and is hereafter called a particle - hole symmetric ( phs ) edge mode . for spatially constant , non - vanishing pair potentials , we have confirmed that @xmath80 for all values of @xmath46 , for a sufficiently large system size where the finite - size discretization of the energy levels is small compared to @xmath81 where @xmath82 is the fermi wavenumber . namely , even deep inside the bcs regime , and even for a small @xmath9 , we find @xmath83 in the thermodynamic limit , which implies no reduction of the oam : @xmath77 holds exactly . a natural question arising here is why @xmath61 remains zero in the bcs regime , despite the quantum phase transition separating it from the bec regime . let us first consider the limit @xmath84 with a fixed @xmath85 , where @xmath86 holds trivially . thus we find @xmath87 in this limit . increasing @xmath33 , @xmath0 and @xmath2 acquire non - zero values . however , as long as there is no gap closing , @xmath87 and thus @xmath77 still hold . this gives a proof for the physical expectation that @xmath77 holds throughout the bec regime , which belong to the non - topological strong - pairing phase . the value of @xmath88 in the bcs regime ( weak - pairing phase ) is more subtle , due to the presence of the quantum phase transition at @xmath89 in the thermodynamic limit . at the quantum phase transition , a gap closing is expected . however , a careful examination reveals that every eigenvalue keeps its sign when @xmath33 is continuously varied from @xmath84 to @xmath90 . in fact , for the system defined on a finite disc , the gap never closes . the gap does approach zero at the quantum critical point , and also inside the bcs regime giving rise to the gapless chiral edge mode , but only in the thermodynamic limit . the change of the eigenvalues when @xmath33 is varied is shown in fig . [ fig : p+ip ] ( d ) . the edge mode appears in the bcs regime , as a set of eigenstates separated from bulk states . although it converges to the linear gapless dispersion , all the eigenvalues corresponding to the edge mode for @xmath91 come off from the upper continuum of bulk eigenstates and remain positive . likewise , all the edge mode eigenvalues for @xmath92 come from the lower continuum and remain negative . since all the eigenvalues depend continuously on @xmath33 , this is the only possible evolution to generate the phs edge mode , under the particle - hole symmetry . although each of @xmath0 and @xmath2 changes from the trivial values @xmath86 , @xmath93 is always satisfied and any correction factor like @xmath94 discussed in the introduction can not arise . we note that this is also true for a physical process where @xmath33 and @xmath75 are simultaneously tuned to keep @xmath2 constant . our argument only relies on the formation of the phs edge mode separated from the bulk eigenstates , which is valid for a sufficiently large system size . the conclusion of our analysis largely agrees with the recent related calculations on @xmath6-wave sf @xcite , but clarify why @xmath11 is exactly given by @xmath95 even for small @xmath96 . lr -wave sf when @xmath97 for ( a ) @xmath98 ( bec regime ) , ( b ) @xmath89 , and ( c ) @xmath99 ( bcs regime ) . ( d ) evolution of the spectrum for fixed @xmath100 as @xmath33 is changed . ] -wave sf when @xmath97 for ( a ) @xmath98 ( bec regime ) , ( b ) @xmath89 , and ( c ) @xmath99 ( bcs regime ) . ( d ) evolution of the spectrum for fixed @xmath100 as @xmath33 is changed . ] cc -wave sf when @xmath97 for ( a ) @xmath98 ( bec regime ) , ( b ) @xmath89 , and ( c ) @xmath99 ( bcs regime ) . ( d ) evolution of the spectrum for fixed @xmath100 as @xmath33 is changed . ] -wave sf when @xmath97 for ( a ) @xmath98 ( bec regime ) , ( b ) @xmath89 , and ( c ) @xmath99 ( bcs regime ) . ( d ) evolution of the spectrum for fixed @xmath100 as @xmath33 is changed . ] next , we move to the @xmath15-wave states for which the spectrum is particle - hole symmetric about @xmath101 . we have numerically confirmed that @xmath102 for all @xmath46 in the bec regime and obtain @xmath103 , i.e. @xmath104 . on the other hand , in the bcs states , it is known that a @xmath15-wave state has two non - degenerate edge modes at one boundary as visualized in fig . [ fig : d+id ] ( a ) @xcite . ccc -wave sf and ( b ) the spectral asymmetry @xmath67 for @xmath105 ( bcs regime ) . ( c ) spectral flow for fixed @xmath106 as @xmath33 is changed . ] -wave sf and ( b ) the spectral asymmetry @xmath67 for @xmath105 ( bcs regime ) . ( c ) spectral flow for fixed @xmath106 as @xmath33 is changed . ] -wave sf and ( b ) the spectral asymmetry @xmath67 for @xmath105 ( bcs regime ) . ( c ) spectral flow for fixed @xmath106 as @xmath33 is changed . ] each edge mode is particle - hole symmetric with the other branch , but not symmetric by itself ; we call them non - phs edge modes . their dispersion relations are given as @xmath107 , where @xmath108 are the `` fermi angular momenta '' where the edge modes cross the zero energy . the phs of the system requires @xmath109 . interestingly , for this spectrum , we find that some of the @xmath67 s become non - vanishing , as shown in fig . [ fig : d+id](b ) : @xmath102 for @xmath110 , @xmath111 , and @xmath101 , while @xmath112 for @xmath113 and @xmath114 for @xmath115 . this result can be understood in terms of the spectral flow staring from the bec regime where @xmath102 . as in the case of the @xmath6-wave sf , @xmath102 remains valid up to the critical point , since the gap remains open in the entire bec regime . at the critical point , the gap in the spectrum is minimal at @xmath101 as shown in fig . [ fig : p+ip ] ( b ) , corresponding to the closing of the gap . however , the gap in a finite - size system does not vanish at @xmath101 . as we move into the bcs regime , two non - phs edge modes develop as in fig . [ fig : d+id ] ( a ) . during this evolution , the fermi angular momenta evolves from @xmath101 to non - vanishing values @xmath116 . this induces spectral flow for the angular momenta @xmath46 in the range @xmath117 , except at @xmath101 . an example of the spectral flow at a fixed value of @xmath46 is shown in fig . [ fig : d+id ] ( c ) : @xmath67 changes sign exactly when a fermi angular momentum passes through this @xmath46 . this picture can be also confirmed in an analytic expression of the edge mode dispersions as functions of @xmath118 , as follows @xcite . let us consider the limit of a large disc radius @xmath37 . the edge of the disc then corresponds to the boundary of a semi - infinite plane , where the momentum parallel to the boundary , @xmath119 , is related to the angular momentum @xmath46 by @xmath120 in the bcs regime @xmath121 , the dispersion of the edge modes are given as @xmath122 where @xmath123 by solving the bdg equation @xcite . the condition @xmath124 then determines the fermi wavevector @xmath125 of the edge modes as @xmath126 this indeed demonstrates that the fermi wavevector @xmath125 evolves from zero to a non - vanishing value , as @xmath33 is increased from the critical point @xmath89 into the bcs regime . this confirms the spectral flow for @xmath127 as found numerically . in fact , spectral flows , and @xmath128 as a consequence , are common properties of the higher - order pairing states with @xmath7 which have non - phs edge modes in the bcs regime . we have also numerically calculated @xmath67 for @xmath129 . as expected , we find @xmath102 for all @xmath46 in the bec regime . in the bcs regime , on the other hand , there appear @xmath4 edge modes : one phs edge mode and @xmath130 non - phs edge modes for an odd @xmath4 , and @xmath4 non - phs edge modes for an even @xmath4 . let @xmath131 be fermi angular momenta for these edge modes , namely the edge modes cross zero energy at @xmath132 . for @xmath133 , we find @xmath134 if @xmath135 and @xmath102 otherwise ; for @xmath136 , @xmath137 if @xmath138 , @xmath139 if @xmath140 or @xmath141 , and @xmath102 otherwise . these results can be naturally understood in terms of the spectral flow starting from @xmath142 deep in the bec regime , as in the case of @xmath20 discussed above . this pattern is expected to persist for any @xmath27 . to summarize , in general chiral sf with @xmath27 , @xmath67 changes by @xmath143 when a non - phs edge mode branch crosses zero energy , while a phs edge mode does not contribute to @xmath67 . in the bcs regime for @xmath7 , non - vanishing @xmath67 implies @xmath144 . as a consequence , @xmath0 is in fact strongly suppressed from the `` full '' value @xmath55 , in the bcs limit @xmath145 . to see this , we evaluate the actual value of @xmath146 using eq . , with the observations made above . by considering the limit of a large disc , the `` fermi angular momenta '' @xmath147 can be written in terms of the fermi wavenumber parallel to the boundary @xmath148 as @xmath149 . within the quasi - classical formulation , which is legitimate for the bcs limit , we find @xmath150 @xcite . thus , in the leading order in @xmath2 and @xmath96 , we obtain @xmath151 since @xmath152 , the oam is evaluated to be @xmath153 in the bcs limit with @xmath154 , where the @xmath155 term represents possible additional contributions which are beyond the quasi - classical approximation . indeed , numerical calculations of the oam give @xmath156 when @xmath157 for @xmath158 in an extended range of the bcs regimes with @xmath159 , supporting the above quasi - classical analysis . therefore , the naive evaluation @xmath3 fails for the chiral sfs in the bcs regime with @xmath7 , even though it gives the correct value for the @xmath6-wave states . that is , @xmath0 is strongly suppressed as if in the naive weak - pairing picture where fermions only near the fermi surface at @xmath160 contribute to @xmath0 , but only for @xmath27 . however , our findings , in particular the stark difference between the @xmath6 and higher - order ( @xmath27 ) pairing cases , make it clear that the suppression can not be understood by any of the arguments found in existing works . our analysis is based on the robustness of the spectral asymmetry @xmath67 , and does not rely on assumptions and approximations used in the eariler papers , such as derivative expansions , which might fail to describe the correct physics especially around boundaries where the edge modes are located @xcite . we emphasize that the well - known topological protection of existence of @xmath4 edge modes is not sufficient for determining the oam , which depends on more detailed structures of the edge modes . in this sense , the oam @xmath0 is a surface - dependent quantity in the bcs regimes . finally , let us discuss why the oam is suppressed for @xmath27 but not for @xmath19 , in terms of the ground - state wave function . a general expression @xcite for the ground state of a bdg hamiltonian is given as @xmath161 , where @xmath162 here @xmath163 is the vacuum for @xmath164 and @xmath165 is a normalization constant , @xmath166 is a linear superposition of @xmath167 , and @xmath168 , @xmath169 are non - negative integers . the ground state of a bdg hamiltonian is often assumed to have a pure exponential form ( @xmath170 in eq . ) , which implies that all the fermions are paired and thus @xmath21 . for @xmath19 , the ground state ( of a sufficiently large system ) is indeed reduced to the pure exponential form , implying the full oam @xmath77 . however , the ground state of a bdg hamiltonian generally takes the form of eq . . a non - vanishing @xmath171 signals the existence of _ unpaired _ fermions , which contribute to the reduction of the oam . in fact , we can derive @xcite the identity @xmath172 which explicitly shows that the unpaired fermions are necessary for the reduction of the oam , which requires @xmath173 . furthermore , the reduction requires a non - vanishing difference @xmath174 . indeed , when @xmath175 , @xmath62 belongs to the eigenvalue @xmath103 . this is because the vacuum @xmath163 belongs to @xmath176 and every creation operator in eq . comes in the pair @xmath177 , which commutes with @xmath178 . comparing eqs . and , we obtain @xmath179 . together with the property @xmath180 , we find the numbers of unpaired fermions explicitly as follows @xcite : for @xmath19 , @xmath181 , namely there is no unpaired fermions as mentioned earlier . for @xmath27 , @xmath182 for @xmath183 , and @xmath184 for @xmath185 . therefore , the unpaired fermions generally carry angular momentum _ opposite _ to the given chirality , and this contribution cancels out with that from the paired fermions , leading to the reduction of @xmath0 from the full value @xmath55 . this depairing effect is associated with the formation of the non - phs edge modes , signifying the fundamental difference between @xmath19 and @xmath27 cases . we thank y. maeno for introducing the problem of the oam in chiral superfluids to us . we are also grateful to e. demler , s. fujimoto , a. furusaki , j. goryo , o. ishikawa , n. kawakami , y. b. kim , t. kita , t. mizushima , s .- j . mao , e .- g . moon , y. nishida , m. sato , k. shiozaki , a. shitade , m. sigrist , h. sumiyoshi , w .- f . tsai , y. tsutsumi , k. ueda , and s .- k . yip for valuable discussions . this work was supported by the `` topological quantum phenomena '' ( no . 25103706 ) grant - in aid for scientific research on innovative areas from the ministry of education , culture , sports , science and technology ( mext ) of japan . 40 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , _ _ ( , ) , ed . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _ ( , ) . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , , , , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , , * * , ( ) . ( ) , ; , , , , , , ( ) , . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) .
we have made a spectroscopical follow - up of 10 pdcs lines of sight including 15 clusters . these pdcs clusters ( lubin et al . 1996 ) are optically selected cluster candidates . we have measured about 700 redshifts in 6 nights at cfh with the mos spectrograph . the expected redshift of these clusters is around 0.4 . two areas of the sky are particularly well covered : 4 lines of sight around 9h and 3 lines of sight around 13h . the sampled redshift range is z=[0.,0.9 ] . the first results of this survey show that more than 60% of the candidate clusters are real gravitationally bounded structures with a velocity dispersion ranging from 600 km / s to 1500 km / s . using the 6 best sampled clusters , we have used the same techniques as were were used for the nearby enacs clusters ( adami et al . 1998 ) . we plot the variation of the galaxy normalized velocity dispersion versus the absolute magnitude and the morphological type ( fig 1 : we assume as a first approximation that the em line galaxies are a mix of early and late spirals and that the only absorption line galaxies are a mix of ellipticals and so ) . the open symbols are the results for nearby clusters ( [email protected] : adami et al . 1998 ) and the filled triangles are for the pdcs clusters ( [email protected] ) . we see that the two distributions are essentially the same : the evolution of the internal dynamics of the clusters between 0.05 and 0.4 seems to be negligible . at [email protected] or [email protected] , the emission line galaxies are an infalling population while the absorption line galaxies seem to be virialized and the galaxies follow the energy equipartition law ( thick line in fig 1 left ) . the conclusion is that the epoch formation of the clusters is probably significantly greater than 0.4 , the clusters continually evolving after with late type galaxies still infalling at low redshifts . the 2 well sampled cop areas ( 9h and 13h ) allow a study of the periodicity along the line of sight . the structures are defined exactly as the enacs ( katgert et al . at 9h , we find a periodicity of 90 @xmath1 2 mpc and at 13h the periodicity is 143 @xmath1 10 mpc . comparing these results with broadhurst et al ( 1990 ) ( 128 mpc in another direction ) , we tentatively conclude that periodicities in the structure distribution are in agreement with the `` web '' representation of the universe and the value of this periodicity depends of the line of sight . adami c. , biviano a. , mazure a. , 1998 , a@xmath2a 331 , 439 broadhurst t. , ellis r. , koo d. , et al . , nature 343 , 726 katgert p. , mazure a. , perea j. , et al . , 1996 , a@xmath2a 310 , 8 lubin l. , postman m. , 1996 , aj 111 , 1795	we present in this paper the first results of the cop survey about the reality of the pdcs clusters , about their velocity dispersions and dynamic and about the periodicity of the structures along the line of sight .
the belle experiment successfully operated for more than a decade until 2010 at the asymmetric - energy @xmath4 collider kekb @xcite . the experiment took data at center - of - mass energies corresponding to several @xmath5s ) resonances ; the total data sample recorded exceeds @xmath6 . here we present recent results based on the full data sample . the belle detector is a large - solid - angle magnetic spectrometer that consists of a silicon vertex detector ( svd ) , a 50-layer central drift chamber ( cdc ) , an array of aerogel threshold cherenkov counters ( acc ) , a barrel - like arrangement of time - of - flight scintillation counters ( tof ) , and an electromagnetic calorimeter comprised of csi(tl ) crystals ( ecl ) located inside a super - conducting solenoid coil that provides a 1.5 t magnetic field . an iron flux - return located outside of the coil is instrumented to detect @xmath7 mesons and to identify muons ( klm ) . the detector is described in detail elsewhere @xcite . in the standard model ( sm ) , the decay @xmath9 proceeds through electroweak and gluonic @xmath10 penguin annihilation processes . these amplitudes are proportional to the small cabibbo - kobayashi - maskawa @xcite matrix element @xmath11 and thus are highly suppressed . the branching fraction has been estimated based on naive qcd factorization @xcite and perturbative qcd @xcite and found to be in the range @xmath12 to @xmath13 . however , the internal loop can also be mediated by non - sm particles such as a charged higgs boson or supersymmetric squarks , and thus the decay is sensitive to new physics ( np ) . it is estimated that such np could enhance the branching fraction to the level of @xmath14 to @xmath15 @xcite . previously , no evidence for this decay has been found . here we describe a recent belle search for this decay with much higher sensitivity than previously that uses the full data set of 711 recorded on the @xmath16(4s ) resonance @xcite . candidate @xmath17 mesons are reconstructed via @xmath18 decays . the @xmath19 invariant mass is required to be in the range @xmath20 $ ] gev/@xmath21 , which corresponds to @xmath22 in resolution around the @xmath17 mass . candidate @xmath23 mesons are identified using a modified beam - energy - constrained mass @xmath24 , and the energy difference @xmath25 , where @xmath26 is the beam energy and @xmath27 and @xmath28 are the momentum and energy , respectively , of the @xmath29 candidate . all quantities are evaluated in the center - of - mass frame . to improve the @xmath30 resolution , the momentum @xmath27 is calculated as @xmath31 , where @xmath32 is the photon momentum , and @xmath33 and @xmath34 are the energy and momentum , respectively , of the @xmath17 candidate . we require that events satisfy @xmath35 and @xmath36 . the signal yield is calculated in a smaller `` signal region '' @xmath37 and @xmath38 . charmless hadronic decays suffer from large backgrounds arising from continuum @xmath39 production . to suppress this background , we use a multivariate analyzer based on a neural network ( nn ) @xcite . the nn generates an output variable @xmath40 , which ranges from @xmath41 for background - like events to @xmath42 for signal - like events . we make a loose requirement @xmath43 and then translate @xmath40 to @xmath44 , defined as @xmath45 $ ] where @xmath46 = 0.3 and @xmath47 = 1.0 . the variable @xmath44 is well - described by a sum of gaussian functions . after the above selections , 961 events remain . the remaining background consists of continuum events and rare charmless @xmath48-decay processes . we obtain the signal yield using an unbinned extended maximum likelihood fit to the observables @xmath30 , @xmath49 , @xmath44 , and @xmath50 . the helicity angle @xmath51 is the angle between the @xmath52 momentum and the opposite of the @xmath23 flight direction in the @xmath17 rest frame . the resulting branching fraction is calculated as @xmath53 where @xmath54 is the signal yield in the signal region , @xmath55 is the number of @xmath56 events , @xmath57 is the signal efficiency as calculated from monte carlo ( mc ) simulation , and @xmath58 @xcite . we find no evidence for this decay and set an upper limit on the branching fraction of @xmath59 at 90% confidence level ( c.l . ) . this limit is almost an order of magnitude lower than the previous most stringent result @xcite . the rare decay @xmath60 involves the quark transition @xmath61 , a flavor changing neutral current that is forbidden at tree level in the sm . higher order sm processes such as penguin or @xmath62 box diagrams allow for such transitions , leading to branching fractions below @xmath63 . various extensions to the sm predict np amplitudes that can interfere with the sm amplitudes and lead to enhanced or suppressed branching fractions and modified angular distributions . belle performed an angular analysis using the decay modes @xmath64 and @xmath65 @xcite . @xmath66 candidates are reconstructed in the channel @xmath67 @xcite . large combinatoric background is suppressed by applying requirements on kinematic variables . large backgrounds also arise from charmonium decays @xmath68 and @xmath69 , in which @xmath70 . to maximize signal efficiency and purity , neural networks are applied sequentially beginning from the end of the decay chain . signal and background yields are determined from an unbinned extended maximum likelihood fit to @xmath30 in bins of @xmath71 . the bin ranges used and the resulting fitted yields are listed in table [ tab : kll_yields ] . in total , @xmath72 signal candidates are obtained for @xmath73 , and @xmath74 candidates for @xmath64 . cccc bin & @xmath71 range in @xmath75 & @xmath76 & @xmath77 + 0 & @xmath78 & @xmath79 & @xmath80 + 1 & @xmath81 & @xmath82 & @xmath83 + 2 & @xmath84 & @xmath85 & @xmath86 + 3 & @xmath87 & @xmath88 & @xmath89 + 4 & @xmath90 & @xmath91 & @xmath92 + we subsequently perform the angular analysis . the decay is kinematically described by three angles ( @xmath93 , @xmath94 , and @xmath17 ) and the invariant mass squared of the lepton pair ( @xmath71 ) . the definitions of the angles and the full angular distribution are described in ref . @xcite . for this fit we require @xmath95 , and the number of signal ( @xmath76 ) and background events ( @xmath77 ) are fixed to the values in table [ tab : kll_yields ] . the angular observables @xmath96 introduced in ref . @xcite contain information about short - distance effects . these observables are considered to be largely free from form - factor uncertainties @xcite , and thus they are a promising place to search for np . as the statistics in this analysis are insufficient to perform a full eight - dimensional fit , a `` folding technique '' is used as described in ref . we determine @xmath97 by performing a three - dimensional unbinned maximum likelihood fit in four bins of @xmath71 using the folded signal probability density function and fixed background yields and shapes . we also fit for the longitudinal polarization @xmath98 and the transverse polarization asymmetry @xmath99 . the fit results are shown in fig . [ fig : kll_proj ] . these results are consistent with sm predictions , although one value of @xmath100 differs by @xmath101 from the sm value . it is notable that this deviation is for the same @xmath71 bin and in the same direction as a similar deviation reported by lhcb @xcite@xcite . results for @xmath100 as compared to sm predictions and to previous measurements by lhcb @xcite@xcite.,scaledwidth=55.0% ] -0.20 in semi - tauonic @xmath23 decays of the type @xmath104 are sensitive probes to search for np . charged higgs bosons , which appear in supersymmetry and other models with at least two higgs doublets , may induce measurable effects in the branching fraction due to the large mass of the @xmath105 lepton . similarly , leptoquarks , which carry both baryon number and lepton number , can also contribute to this process . the ratio of branching fractions @xmath106 is typically measured instead of the absolute branching fraction @xmath107 to reduce systematic uncertainties arising from reconstruction efficiencies , form factors , and the ckm matrix element @xmath108 . standard model calculations predict @xmath109 @xcite and @xmath110 @xcite@xcite . semi - tauonic @xmath23 decays were first observed by belle @xcite , with subsequent studies reported by belle @xcite@xcite , @xcite , and lhcb @xcite . the world average values @xcite are @xmath111 and @xmath112 , which exceed the sm predictions by @xmath113 and @xmath114 , respectively . thus far , measurements of @xmath115 at belle and have been performed using either a hadronic @xcite@xcite or an inclusive @xcite@xcite tagging method . here we report the first measurement of @xmath116 using a semileptonic tagging method . we reconstruct signal events in modes in which one @xmath23 decays as @xmath117 and the the other @xmath23 decays as @xmath102 , @xmath118 . to reconstruct normalization events corresponding to the denominator in eq . ( [ eqn : rdstar ] ) , we require that both @xmath23 mesons decay to @xmath119 . signal and normalization events are identified using a neural network . the dominant background contribution arises from events with falsely reconstructed @xmath120 mesons . to separate signal and normalization events from backgrounds , we use the energy variable @xmath121 , which is defined as the sum of the energies of neutral clusters detected in the electromagnetic calorimeter ( ecl ) that are not associated with reconstructed particles . we extract the signal and normalization yields using a two - dimensional extended maximum - likelihood fit to the neural network output nn and @xmath121 . the resulting yields of signal and normalization events are @xmath122 and @xmath123 , respectively . the ratio @xmath116 is @xmath124 , where the first and second errors correspond to statistical and systematic uncertainties , respectively . our measurement is @xmath125 higher than the sm prediction . we investigate the compatibility of the data with a type ii two - higgs doublet model ( 2hdm ) @xcite and with @xmath126-type and @xmath127-type leptoquark models @xcite . we find that the data allows for additional contributions from 2hdm scalar operators and also vector operators , but contributions from a tensor operator with @xmath128 , an @xmath129-type leptoquark model with @xmath130 , or an @xmath131-type leptoquark model with @xmath132 are disfavored . the two - body decays @xmath134 , where @xmath135 is either a charged @xmath136 or @xmath137 , have all been observed @xcite . however , the neutral modes @xmath138 , @xmath139 , and @xmath140 have not . the decay @xmath141 is of particular interest because the branching fraction is predicted to be large : @xmath142 @xcite . the presence of non - sm particles or couplings could measurably affect this value @xcite . the current upper limit , @xmath143 at 90% c.l . , was set by belle using only @xmath144 of data recorded at the @xmath3 @xcite . here we update that result using the full data set of @xmath145 @xcite . at the @xmath3 , @xmath146 pairs are produced in three decay channels : @xmath146 , @xmath147 or @xmath148 , and @xmath149 . the latter two channels dominate , with production fractions of @xmath150 and @xmath151% @xcite . candidate @xmath152 mesons are reconstructed via the decay @xmath153 using a neural network technique . to suppress background arising from @xmath154 production , a second nn is used . the output of the latter ( @xmath40 ) ranges from @xmath41 for background - like events to @xmath42 for signal - like events . we make a loose requirement @xmath155 and then translate @xmath40 to the variable @xmath44 as described in section [ sec : rho_gamma ] . we measure the signal yield by performing an unbinned extended maximum likelihood fit to variables @xmath30 , @xmath49 , and @xmath156 . the results are @xmath157 signal events and @xmath158 background events . projections of the fit result are shown in fig . [ fig : bskk - fig2 ] . the branching fraction is calculated as @xmath159 where @xmath76 is the fitted signal yield ; @xmath160 is the number of @xmath146 events @xcite ; @xmath161 is the branching fraction for @xmath153 @xcite ; and @xmath162 is the reconstruction efficiency as determined from mc simulation . the factor 0.50 accounts for the 50% probability for @xmath163 ( since @xmath140 is @xmath164 even ) . inserting these values gives @xmath165 , where the first uncertainty is statistical , the second is systematic , and the third reflects the uncertainty due to the total number of @xmath146 pairs . this result is in good agreement with the sm prediction @xcite . projections of the fit for @xmath133 : ( a ) @xmath30 for @xmath166 and @xmath167 ; ( b ) @xmath49 for @xmath168 and @xmath167 ; and ( c ) @xmath156 for @xmath169 and @xmath170 . the points with error bars are data , the ( green ) dashed curves show the signal , ( magenta ) dotted curves show the continuum background , and ( blue ) solid curves show the total . , title="fig:",scaledwidth=32.0% ] projections of the fit for @xmath133 : ( a ) @xmath30 for @xmath166 and @xmath167 ; ( b ) @xmath49 for @xmath168 and @xmath167 ; and ( c ) @xmath156 for @xmath169 and @xmath170 . the points with error bars are data , the ( green ) dashed curves show the signal , ( magenta ) dotted curves show the continuum background , and ( blue ) solid curves show the total . , title="fig:",scaledwidth=32.0% ] projections of the fit for @xmath133 : ( a ) @xmath30 for @xmath166 and @xmath167 ; ( b ) @xmath49 for @xmath168 and @xmath167 ; and ( c ) @xmath156 for @xmath169 and @xmath170 . the points with error bars are data , the ( green ) dashed curves show the signal , ( magenta ) dotted curves show the continuum background , and ( blue ) solid curves show the total . , title="fig:",scaledwidth=32.0% ] -0.20 in the signal significance is calculated as @xmath171 , where @xmath172 is the likelihood value when the signal yield is fixed to zero , and @xmath173 is the likelihood value of the nominal fit . systematic uncertainties are included in the significance by convolving the likelihood function with a gaussian function whose width is equal to the systematic uncertainty associated with the signal yield . we obtain a signal significance of @xmath174 , and thus our measurement constitutes the first observation of this decay . @xmath164 asymmetries in the charm sector are a promising area in which to search for new physics . the asymmetry ( @xmath176 ) of many two - body @xmath177 decays has been measured @xcite , but all results are consistent with zero . here we present a first search for a asymmetry in radiative @xmath175 decays , where @xmath178 is a neutral vector meson @xmath17 , @xmath179 , or @xmath180 . within the sm , such asymmetries are predicted to be small : @xmath181 . however , theoretical studies @xcite@xcite indicate that extensions to the sm with chromomagnetic dipole operators can increase @xmath176 up to a few percent . several @xmath175 decay modes have been measured by belle and @xcite@xcite . the current world average branching fractions are @xmath182 and @xmath183 @xcite . the @xmath184 mode has not yet been observed , and the current upper limit on the branching fraction is @xmath185 at 90% c.l . @xcite . here we present new measurements of all three modes using the full belle data set . our results include the first measurement of the asymmetries . for our analysis , the @xmath186 is required to originate from @xmath187 decays ; this tags the @xmath186 flavor and suppresses combinatoric background . the vector mesons in the final state are reconstructed via @xmath188 , @xmath189 , and @xmath190 . all branching fractions and @xmath176 are determined by normalizing to decay channels @xmath191 for @xmath192 , @xmath193 for @xmath194 , and @xmath195 for @xmath196 . the branching fractions are calculated as @xmath197 where @xmath198 and @xmath199 are the branching fraction and fitted yield , respectively , of signal or normalization modes , and @xmath200 is the reconstruction efficiency . for @xmath201 , the world - average value @xcite is used . the extracted raw asymmetry @xmath202 has several contributions : @xmath203 , where @xmath204 is the forward - backward production asymmetry and @xmath205 is the detection asymmetry between positively and negatively charged particles . both of these asymmetries are eliminated by measuring @xmath206 of a signal mode relative to that of its normalization mode , which has the same final - state charged particles . the @xmath164 asymmetry of a signal mode is then @xmath207 , where @xmath208 is the pdg value of the @xmath164 asymmetry for the normalization mode @xcite . to extract the signal yields and @xmath164 asymmetries , we perform a simultaneous two - dimensional unbinned extended maximum likelihood fit of the @xmath186 and @xmath209 samples . the fit variables are the invariant mass of the @xmath186 and the cosine of the helicity angle @xmath210 , defined as the angle between the @xmath186 and the positively or negatively charged hadron in the rest frame of the @xmath178 meson . for @xmath186 candidates we use the @xmath52/@xmath211/@xmath212 for @xmath17/@xmath213/@xmath180 decays , and we use the oppositely charged particles for @xmath209 candidates . the fitted signal yields are @xmath214 , @xmath215 , and @xmath216 . the resulting branching fractions are @xmath217 where the first uncertainty is statistical and the second is systematic . the result for @xmath218 is improved with respect to the previous result @xcite . the result for @xmath219 is 3.3@xmath220 higher than the result @xcite . the result for @xmath221 is close to that for @xmath218 , which is consistent with theoretical expectations . the significance is calculated as @xmath222 as described in section [ sec : bskk ] . systematic uncertainties are included in the significance by convolving the likelihood function with a gaussian whose width is equal to the systematic uncertainty associated with the signal yield . the significance for @xmath221 is @xmath223 , and thus our measurement constitutes the first observation of this decay . the results for @xmath176 are @xmath224 no asymmetry is seen in any of these radiative decays . the authors thank the beach 2016 organizers for a well - run workshop and excellent hospitality . this research is supported by the u.s . department of energy .	we review recent results from the belle experiment , which took data at the kekb asymmetric - energy @xmath0 collider in japan . the experiment recorded about 1000 fb@xmath1 of data running mainly at the @xmath2 and @xmath3 resonances . the results presented here are obtained from the full data set .
the objective of this paper is to study the korn - poincar inequality : @xmath14 under the tangential boundary conditions : @xmath15 in domains @xmath1 that are shells of small thickness of order @xmath2 , around an arbitrary smooth and closed hypersurface @xmath3 in @xmath4 . by @xmath16 we denote the symmetric part of the gradient @xmath6 . korn s inequality was discovered in the early @xmath17th century , in the context of the boundary value problem of linear elastostatics @xcite . there is by now an extensive literature on the subject , relating to various contexts and various boundary conditions ( see for example a review @xcite , and the references therein ) . if ( [ intro_tan ] ) is replaced by @xmath18 on @xmath19 , one can easily prove that @xmath20 , and so ( [ intro_korn ] ) follows by the poincar inequality . in the absence of this boundary condition , or with its weaker versions , the bound ( [ intro_korn ] ) requires an extra criterion . roughly speaking , it serves to eliminate pure rotations , when @xmath21 but @xmath22 . in particular , ( [ intro_korn ] ) holds for all @xmath23 vector fields @xmath10 with ( [ intro_tan ] ) , which are @xmath24- orthogonal to the space of those linear vector fields on @xmath1 with skew - symmetric gradient that are themselves tangent on the boundary . we are interested in the behaviour of the constant @xmath8 , as @xmath9 . it turns out that in general , @xmath8 may blow up , even if @xmath1 are not rotationally symmetric ( that is when the aforementioned spaces are trivial ) . the correct way of looking at this problem is to consider the asymptotic inequality as @xmath9 , i.e. the related korn inequality on @xmath3 ( see also @xcite ) : @xmath25 this inequality holds true for all tangent vector fields @xmath26 on @xmath3 , which are @xmath24-orthogonal to the the space of killing fields on @xmath3 . a killing field is defined to be a smooth tangent vector field which generates a one - parameter family of isometries on @xmath3 . it is well known that the space of killing fields on a given surface is a finite dimensional lie algebra . an equivalent characterisation of killing fields on @xmath3 is : @xmath27 in this paper , we first notice that any @xmath26 satisfying ( [ intro_tre ] ) admits a family of extensions @xmath28 , such that the boundary conditions ( [ intro_tan ] ) hold and so that the ratio @xmath29 goes to infinity as @xmath9 . this construction turns out to be the worst case scenario for the possible blow - up of @xmath8 . our main results state that the constants @xmath8 remain uniformly bounded for vector fields @xmath10 inside any family of cones ( with angle @xmath11 , uniform in @xmath2 ) around the orthogonal complement of the space of extensions of killing fields on @xmath3 . our main motivation in this work has been its application to dynamics of navier - stokes equations in thin @xmath30-dimensional domains . thin domains are encountered in many problems in solid or fluid mechanics . for example , in ocean dynamics , one is dealing with the fluid regions which are thin compared to the horizontal length scales . other examples include lubrication , meteorology , blood circulation etc . ; they are a part of a broader study of the behaviour of various pdes on thin @xmath31-dimensional domains , where @xmath32 ( for a review see @xcite ) . the study of the global existence and asymptotic properties of solutions to the navier - stokes equations in thin @xmath30d domains began with raugel and sell in @xcite . they proved global existence of strong solutions for large initial data and in presence of large forcing , for the sufficiently thin @xmath30d product domain @xmath33 , with the boundary conditions either purely periodical or combined periodic - dirichlet . further generalisations to other boundary conditions followed ( see the references in @xcite ) . towards analysing thin domains other than simple product domains , iftimie , raugel and sell @xcite treated domains of the type : @xmath34 , with the mixed boundary conditions : periodic on the lateral boundary and the navier boundary conditions : @xmath35 on the top and on the bottom . the korn inequality arises naturally when one considers the incompressible flow subject to ( [ intro_navier ] ) , for the following reason ( see @xcite ) . in order to define the stokes operator relevant to the boundary conditions ( [ intro_navier ] ) , one uses the symmetric bilinear form @xmath36 rather than the usual @xmath37 . hence , the energy methods give suitable bounds for @xmath38 , for a solution flow @xmath12 in @xmath1 . on the other hand , in order to establish compactness while studying the limit problem as @xmath39 , one needs bounds for the @xmath40 norm of @xmath12 , with constants independent of @xmath2 . the inequality ( [ intro_korn ] ) ( with uniform constants @xmath8 ) provides thus a necessary uniform equivalence of the two norms @xmath41 and @xmath42 on @xmath1 . in this spirit , it is hoped that we can apply the result of this paper to study the dynamics of the navier - stokes equations , under the navier boundary conditions , in thin shells with various geometries of the reference surface @xmath3 and of the boundaries of @xmath1 . starting with the original papers of korn @xcite , korn s inequality has also been widely used as a basic tool for the existence of solutions of the linearised displacement - traction equations in elasticity @xcite . in this context , for a given displacement vector field @xmath10 , the matrix field @xmath5 is the linearised strain , which measures the pointwise deviation of the deformation @xmath43 from a rigid motion , up to the first order terms in @xmath44 . hence , korn s inequality can be interpreted as a rigidity estimate for small displacement deformations : they are @xmath40 close to @xmath45 , by the error given in the right hand side of ( [ intro_korn ] ) . a nonlinear version of this rigidity estimate , obtained recently in @xcite , has been extensively applied to problems in nonlinear elasticity and plate theories ( see eg @xcite ) . earlier , korn s inequalites in thin neighbourhoods of flat surfaces have been discussed in series of papers by kohn and vogelius @xcite . they derive an estimate which degenerates as @xmath39 for clamped boundary conditions at the side of the plate . an analogous result in our settting is given in theorem [ th_very_weak ] below . * acknowledgments . * was partially supported by the nsf grant dms-0707275 . let @xmath3 be a smooth , compact hypersurface ( a boundaryless manifold of co - dimension @xmath46 ) in @xmath4 . consider a family @xmath47 of thin shells around @xmath3 : @xmath48 given by a family of smooth positive functions @xmath49 . we will use the following notation : @xmath50 for the outward unit normal to @xmath19 , @xmath51 for the outward unit normal to @xmath3 ( seen as the boundary of some bounded domain in @xmath4 ) , @xmath52 for the tangent space to @xmath3 at a given @xmath53 . the projection onto @xmath3 along @xmath54 will be denoted by @xmath55 , so that : @xmath56 the standard korn inequality ( see theorem [ korn_standard ] in appendix a ) on bounded lipschitz domains implies that for each @xmath57 satisfying the orthogonality condition : @xmath58 one has : @xmath59 and the constant @xmath8 depends only on the domain @xmath1 , but not on @xmath10 . here , @xmath60 stands for the linear space of all @xmath61 skew - symmetric matrices : @xmath62 while by @xmath5 we mean the symmetric part of @xmath6 : @xmath63 the same result is true for @xmath10 satisfying additionally : @xmath64 when in ( [ ort_lin ] ) we take only linear functions @xmath65 ; with skew - symmetric gradient , and satisfying the same boundary condition as @xmath10 : @xmath66 the standard proof by contradiction ( see theorem [ naive_korn ] in appendix a ) shows that the constant @xmath8 in ( [ korn_nonunif ] ) again does not depend on @xmath10 but it may depend on the geometry of @xmath1 . in particular , as shown in the example in section [ example ] , @xmath8 may indeed converge to infinity as the thickness of @xmath1 ( that is @xmath67 ) converges to @xmath68 . our goal is to investigate the behaviour of @xmath8 in two frameworks , relating to the following hypotheses : * for some positive constants @xmath69 and @xmath70 , and all small @xmath71 there holds : @xmath72 * for some smooth positive functions @xmath73 , there holds : @xmath74 notice that clearly * ( h2 ) * implies * ( h1 ) * with , for example : @xmath75 , @xmath76 , @xmath77 . before stating our main results , we need to recall the notion of a killing vector field . the lie algebra of smooth killing vector fields on @xmath3 will be denoted by @xmath78 . that is , @xmath79 if and only if : * @xmath80 is smooth and @xmath81 for every @xmath53 , * @xmath82 for every @xmath53 and every @xmath83 . condition ( ii ) implies that @xmath84 recall that killing vector fields are infinitesimal generators of isometries on @xmath3 , in the sense that if @xmath85 is the flow generated by @xmath26 : @xmath86 then for every fixed @xmath87 the map @xmath88 is an isometry . it is known that the linear space @xmath78 has finite dimension @xcite . we recall this fact in appendix c. given positive smooth functions @xmath73 , define : @xmath89 our main results are the following : [ th1 ] assume * ( h1 ) * and let @xmath90 be any constant . then , for all @xmath71 sufficiently small and all @xmath57 satisfying one of the following tangency conditions : @xmath91 or : @xmath92 together with : @xmath93 there holds : @xmath94 where @xmath95 is independent of @xmath10 and of @xmath2 . [ th2 ] assume * ( h2 ) * and let @xmath90 be any constant . then for all @xmath71 sufficiently small and all @xmath57 satisfying @xmath96 on @xmath19 and : @xmath97 there holds ( [ korn ] ) with @xmath95 independent of @xmath10 and of @xmath2 . the example constructed in section [ example ] shows that conditions ( [ f1 ] ) ( or ( [ f2 ] ) ) are necessary for the bound ( [ korn ] ) . in particular , any killing field @xmath26 on @xmath3 can be extended to a vector field @xmath98 on @xmath1 , satisfying the boundary condition and such that @xmath99 but @xmath100 . hence , if one naively assumes that @xmath10 satisfies the angle condition only with the space of generators of appropriate rotations on @xmath3 , rather than the whole @xmath78 , the constant @xmath8 has a blow - up rate of at least @xmath101 , as @xmath9 . the following theorem shows that this is the actual blow - up rate , under the abovementioned conditions . more precisely , define : @xmath102 @xmath103 [ th_very_weak ] let @xmath90 be any constant . then , for all @xmath2 sufficiently small and all @xmath57 , there holds : @xmath104 in any of the following situations : * * ( h1 ) * holds , @xmath105 on @xmath106 or @xmath105 on @xmath107 , and : @xmath108 * * ( h2 ) * holds , @xmath105 on @xmath19 , and : @xmath109 notice that ( i ) is implied by the hypotheses of theorem [ th1 ] and ( ii ) by the hypotheses of theorem [ th2 ] , as the spaces @xmath110 and @xmath111 are contained in @xmath78 or @xmath112 , respectively . the bound ( [ very_weak ] ) was obtained also in @xcite , but in a different context of thin plates with clamped boundary conditions and rapidly varying thickness . conditions ( [ f1 ] ) and ( [ f2 ] ) may be understood in the following way : the cosine of the angle ( in @xmath113 ) between @xmath10 and its projection onto the linear space @xmath114 of ` trivial ' extensions @xmath115 of certain killing fields @xmath116 ( or @xmath117 ) should be smaller than @xmath118 . equivalently , one considers vector fields @xmath119 , which for a given constant @xmath120 ( related to @xmath118 through : @xmath121 ) satisfy : @xmath122 that is , the distance of @xmath10 from the space @xmath123 controls ( uniformly ) the full norm @xmath124 . by theorems [ th1 ] and [ th2 ] , inside each closed cone around @xmath125 , of fixed angle @xmath126 in @xmath113 , the bound ( [ korn ] ) holds , with a constant @xmath95 , that is uniform in @xmath10 and @xmath2 . one could therefore argue that @xmath123 is the kernel for the uniform korn - poincar inequality , in the same manner as the linear maps @xmath127 with skew gradients @xmath128 constitute the kernel for the standard korn inequality ( [ ort_lin ] ) , ( [ korn_nonunif ] ) . this is not exactly the case , as the uniform korn inequality is true for the extensions @xmath115 ( see remark [ rem_trivial_ext ] ) . the role of the aforementioned kernel is played by the space @xmath129 of other , smart extensions @xmath98 of the killing fields @xmath26 ( see the formula ( [ 3_def ] ) ) . still , with @xmath115 replaced by @xmath98 in ( [ f1 ] ) or ( [ f2 ] ) , both theorems [ th1 ] and [ th2 ] remain true . indeed , notice that the spaces @xmath123 and @xmath129 are asymptotically tangent at @xmath130 : @xmath131 hence , if @xmath132 for some @xmath90 , then @xmath133 , and the angle conditions in main theorems hold , with another @xmath90 , when @xmath2 is sufficiently small . thus , the fact that we chose to work with trivial extensions , in @xmath123 ( giving a simpler condition ) , instead of the real kernel @xmath129 , is not restrictive . in the particular case when @xmath19 is parallel to @xmath3 , say @xmath134 , we have @xmath135 @xmath136 if @xmath137 then @xmath138 is tangent to @xmath3 and , as shown in appendix a ( theorem [ lemma_rotations ] ) it generates a rotation on @xmath3 . actually , one has : @xmath139 and so by the preceding comment we see that the condition ( [ ort_lin ] ) is asymptotically contained in ( [ f1 ] ) ( or ( [ f2 ] ) ) . a natural question is whether @xmath78 may contain other vector fields than the restrictions of generators of rigid motions on the whole @xmath4 . this is clearly the case when @xmath140 : any tangent vector field of constant length is a killing field . the same question for higher dimensions and even for @xmath141 and general ( nonconvex ) hypersurfaces is , to our knowledge , still open . it is closely related to another open problem of whether the class of rotationally symmetric surfaces is closed under intrinsic isometries . a further related question is whether every intrinsic isometry on @xmath3 is actually a restriction of some isometry of @xmath142 . when @xmath3 is convex , it is well known that it is the case , while for non - convex surfaces it is false . the answer to the same question , formulated for @xmath46-parameter families of isometries is not known ( see @xcite vol . 5 ) . * an outline of proofs of theorems [ th1 ] and [ th2 ] . * the general strategy is as follows . suppose that @xmath5 is small . the main idea is to study the map @xmath143 which is obtained by averaging @xmath10 in the normal direction : @xmath144 by the boundary condition , one has @xmath145 , i.e. @xmath146 is almost tangential to @xmath3 . moreover , @xmath147 is essentially bounded by the average of @xmath5 . hence if @xmath5 is small , by korn s inequality on surfaces , the field @xmath146 must be close to a killing field @xmath26 . if @xmath26 is not small , we will get a contradiction to the angle condition ( [ f1 ] ) or ( [ f2 ] ) . if @xmath26 is small then we get good estimates for @xmath146 and ultimately for @xmath10 . more precisely , the proof proceeds as follows . first ( see theorem [ approx ] ) , an application of korn s inequality to cylinders of size @xmath2 and an interpolation argument yield a smooth field @xmath148 such that : @xmath149 from this we deduce ( see lemma [ lem2 ] ) : @xmath150 where @xmath151 for all tangent fields @xmath152 and @xmath153 . using the boundary conditions it is easy to show that ( see lemma [ lem1 ] ) : @xmath154 it is thus natural to study the tangent field : @xmath155 now korn s inequality on @xmath3 implies that there exists a killing field @xmath26 such that : @xmath156 by the angle condition , @xmath26 must be small in @xmath157 , and hence in @xmath158 since the killing fields form a finite dimensional space . thus , @xmath159 is controlled , and by ( [ d ] ) @xmath160 is also controlled . now the crucial step is to combine ( [ b ] ) and ( [ c ] ) to deduce that : @xmath161 ( see lemma [ lem3 ] ) . from ( [ e ] ) and ( [ d ] ) we obtain control on @xmath162 . by ( [ c ] ) this controls @xmath163 , hence @xmath164 , and finally ( [ a ] ) gives the estimate for @xmath6 . the actual argument is by contradiction , assuming that @xmath165 and @xmath166 ( see section [ section_endproof ] ) . above and in all subsequent proofs , @xmath95 denotes an arbitrary positive constant , depending on the geometry of @xmath3 and constants @xmath167 in * ( h1 ) * or the functions @xmath168 in * ( h2). the constant @xmath95 may also depend on the choice of @xmath118 , but it is always independent of @xmath10 and @xmath2 . let @xmath73 be some positive and smooth functions , and let @xmath169 assume that on @xmath3 there exists a nonzero killing vector field @xmath26 such that : @xmath170 we are going to construct a family @xmath171 satisfying the boundary condition @xmath172 for which the uniform bound ( [ korn ] ) is not valid ( after we take @xmath173 ) . by @xmath174 we denote the shape operator on @xmath3 , that is , the ( tangential ) gradient of @xmath175 . for all @xmath53 and all @xmath176 define : @xmath177 by ( [ ex_uno ] ) we easily obtain : @xmath178 which means that each @xmath98 is a linear interpolation between the push - forward of the vector field @xmath26 from @xmath3 onto the external part @xmath106 of the boundary of @xmath1 and the other push - forward onto the internal part @xmath107 of @xmath19 ( see figure [ ex_fig ] ) . indeed , the derivative of the map : @xmath179 is given through : @xmath180 in particular , we see that ( [ ex_bd ] ) holds . [ ex_fig ] write now @xmath181 , with : @xmath182 we wish to estimate the order of different coefficients in @xmath183 and @xmath184 . for every @xmath185 , @xmath53 , there holds : @xmath186 observe that : @xmath187 because @xmath188 and the symmetric form @xmath189 commutes with @xmath190 . likewise : @xmath191 to estimate @xmath192 , for @xmath193 , notice that : @xmath194 because @xmath195 . above and in the sequel , @xmath95 denotes any positive constant independent of @xmath2 . since @xmath196 , by ( [ kill ] ) and ( [ ex_calc_1 ] ) we obtain : @xmath197 and by ( [ ex_calc_2 ] ) , ( [ ex_calc_3 ] ) and ( [ ex_calc_4 ] ) : @xmath198 for every @xmath199 . hence : @xmath200 on the other hand , inspecting the terms in ( [ ex_calc_1 ] ) and recalling that @xmath201 ( and therefore @xmath202 as well ) we see that : @xmath203 the two last inequalities imply that the uniform bound ( [ korn ] ) is not valid , without the restriction ( [ f2 ] ) . even if @xmath3 has no rotational symmetry , the constants @xmath8 in ( [ korn_nonunif ] ) become unbounded as @xmath9 . [ rem_trivial_ext ] our construction ( [ 3_def ] ) is crucial for the counterexample to work . indeed , one can not simply take the trivial extensions @xmath204 to obtain the blow - up of @xmath8 . the reason is that , for any @xmath83 , one has : @xmath205 and thus both @xmath206 and @xmath207 are of the order @xmath208 . hence : @xmath209 where @xmath95 denotes , as usual , any uniform constant . in this section we construct a smooth function @xmath164 with skew - symmetric matrix values , approximating @xmath6 on @xmath1 with the error @xmath210 . this function will play a crucial role in future analysis . the construction relies on appendix b , where the constant in korn s inequality on a fixed , star - shaped domain is proved to depend only on the ratio of the diameters of the domain and the lipschitz constant of its boundary ( theorem [ app_main_korn ] ) . we apply this estimate locally and then use a mollification argument as in @xcite . as always , @xmath95 denotes any uniform constant , independent of @xmath10 and @xmath2 . [ approx ] assume ( h1). for every @xmath211 there exists a smooth map @xmath148 such that : @xmath212 * @xmath213 * 1 . * for @xmath53 consider balls in @xmath3 and cylinders in @xmath1 defined by : @xmath214 the main observation is that an application of korn s inequality on @xmath215 ( see theorem [ app_main_korn ] ) yields a skew - symmetric matrix @xmath216 such that : @xmath217 indeed , recalling the assumption * ( h1 ) * we see that for @xmath2 sufficiently small , @xmath215 are star - shaped with respect to @xmath218 and that both the lipschitz constants of their boundaries and the ratios of their diameters have common bounds . our goal is to replace @xmath219 by a matrix @xmath220 which depends smoothly on @xmath218 . this will allow us to replace @xmath219 by @xmath221 in ( [ 1.uno ] ) . the desired estimate on @xmath1 then follows by summing over a suitable family of cylinders . the smoothness of @xmath164 will also play essential role in the key estimate in lemma [ lem3 ] . * 2 . * to define @xmath220 consider a cut - off function @xmath222 , with @xmath223 , @xmath224 constant in a neighbourhood of @xmath68 , and @xmath225 . for each @xmath53 define : @xmath226 then @xmath227 for @xmath228 and : @xmath229 define @xmath220 as the average : @xmath230 where @xmath231 denotes the skew - symmetric part of a given matrix @xmath232 . since @xmath233 , we have : @xmath234 and by the cauchy - schwarz inequality , noting that @xmath235 we obtain : @xmath236 to estimate the derivative of @xmath164 we use that : @xmath237 thus : @xmath238 and by ( [ 1.uno ] ) : @xmath239 similarly , we get for all @xmath240 : @xmath241 where @xmath242 . from this , by the fundamental theorem of calculus : @xmath243 in combination with ( [ 1.uno ] ) and ( [ 1.due ] ) this yields : @xmath244 now cover @xmath1 with a family @xmath245 so that the covering number of the family @xmath246 is independent of @xmath2 . summing ( [ 1.cinque ] ) over @xmath247 proves ( i ) . finally , integrating ( [ 1.quattro ] ) on @xmath248 we get ; @xmath249 and using the same covering as before we obtain ( ii ) . following the same argument , we will prove a uniform poincar inequality in thin domains - see theorem [ th_uniform_poincare ] in appendix d. let @xmath250 be the average of @xmath10 in the normal direction : @xmath251 in this section we will establish four useful estimates on various components of @xmath146 and their derivatives . the first estimate , on @xmath252 is a natural extension of the previous theorem [ approx ] : [ lem2 ] assume * ( h1). for every @xmath211 there holds : @xmath253 where the subscript @xmath254 refers to the tangential components of the appropriate matrix valued function , that is : @xmath255 and @xmath256 for all @xmath53 and @xmath185 . through a direct calculation one checks that for every @xmath53 and @xmath83 there holds : @xmath257 and : @xmath258 hence , by theorem [ approx ] ( i ) : @xmath259 in order to estimate the normal part @xmath146 , we will use the following simple bounds : [ trivial ] recall that @xmath260 , with : @xmath261 if * ( h1 ) * holds then @xmath262 for all @xmath263 and @xmath264 for all @xmath265 . * if * ( h2 ) * holds then : @xmath266 let now @xmath57 . * @xmath267 for all @xmath199 . * if * ( h1 ) * holds and @xmath105 on @xmath106 , then : @xmath268 * if * ( h2 ) * holds and @xmath105 on @xmath19 : @xmath269 \(i ) is obvious . to prove ( ii ) observe , for example , that on @xmath106 the normal @xmath270 is parallel to @xmath271 , where @xmath272 . normalising this vector we conclude the second inequality in ( ii ) . the first one follows in the same manner . \(iii ) follows from : @xmath273 . to prove ( iv ) , use ( i ) and the trace theorem in appendix e : @xmath274 for ( v ) , use ( ii ) , ( iii ) and theorem [ th_uniform_trace ] : @xmath275 [ lem1 ] assume * ( h1 ) * and let @xmath211 satisfy @xmath276 on @xmath106 . then : @xmath277 by lemma [ trivial ] ( iv ) and ( i ) , for every @xmath278 we obtain : @xmath279 hence by theorem [ th_uniform_trace ] : @xmath280 the next , key estimate , is on the gradient of @xmath281 . it is obtained using the divergence theorem on the surface @xmath3 : [ lem3 ] assume * ( h1 ) * and let @xmath211 satisfy @xmath276 on @xmath106 . then : @xmath282 first note that @xmath283 , since @xmath284 and @xmath285 . to prove the desired estimate we use the hilbert space identity : @xmath286 with @xmath287 and @xmath288 . integration by parts shows that : @xmath289 where we applied the divergence theorem , theorem [ approx ] and lemma [ lem1 ] . on the other hand @xmath290 , so by lemma [ lem2 ] : @xmath291 combining ( [ est3 ] ) and ( [ est4 ] ) proves the result . finally , in presence of the stronger condition * ( h2 ) * , we have an additional bound : [ ass_h2 ] assume * ( h2 ) * and let @xmath211 , @xmath292 on @xmath19 . then : @xmath293 we have : @xmath294 the last inequality follows from lemma [ trivial ] ( v ) and from an easy bound : @xmath295 . in this section we will prove the uniform korn s estimate : @xmath296 under the angle constraints ( [ f1 ] ) or ( [ f2 ] ) . we argue by contradiction ; assume thus that ( [ korn_proof ] ) is not valid , for any uniform constant @xmath95 . hence , there exist sequences @xmath297 and @xmath298 ( for simplicity we will write @xmath2 instead of @xmath299 ) such that the assumptions of theorem [ th1 ] or [ th2 ] are satisfied , but : @xmath300 for the proof of theorem [ th1 ] we will assume that @xmath301 on @xmath106 . the case of the tangency condition on @xmath107 is proven exactly the same . notice that ( [ contra ] ) immediately gives , through lemmas [ lem1 ] , [ lem2 ] and [ lem3 ] , that : @xmath302 also , lemma [ ass_h2 ] implies that under the assumption * ( h2 ) * : @xmath303 where we used that the sequence @xmath304 is bounded in @xmath305 , again in view of ( [ contra ] ) . a contradiction will be derived in several steps . in particular , the tangential component of the average @xmath146 : @xmath306 will be estimated using the korn inequality on hypersurfaces ( see appendix c ) . the conditions ( [ f1 ] ) and ( [ f2 ] ) assumed in theorems [ th1 ] and [ th2 ] will be used in full ( not just for rotations as in theorem [ th_very_weak ] ) . * proof of theorems [ th1 ] and [ th2 ] . * * 1 . * applying theorem [ th_surface_korn ] to each tangent vector field @xmath307 , we obtain a sequence @xmath308 such that : @xmath309 for every @xmath53 and @xmath83 there holds : @xmath310 as @xmath311 . thus , by ( [ sm1 ] ) : @xmath312 therefore : @xmath313 let @xmath314 be the orthogonal projection ( with respect to the @xmath157 norm ) of the space @xmath78 onto its subspace @xmath315 , which we take to be the whole @xmath78 in case of theorem [ th1 ] and @xmath316 in case of theorem [ th2 ] . call @xmath317 and @xmath318 . in both cases ( [ dist_angle ] ) implies : @xmath319 we now prove that : @xmath320 in case of theorem [ th1 ] , when @xmath321 , ( [ nove ] ) is trivial , so we concentrate on the case of theorem [ th2 ] . notice that then , ( [ sm3 ] ) and ( [ due.5 ] ) yield : @xmath322 since all norms in the finitely dimensional space @xmath323 are equivalent , we have : @xmath324 indeed , the right hand side of ( [ dieci ] ) provides a norm on the space in question . now , ( [ h2_case ] ) and ( [ dieci ] ) clearly imply ( [ nove ] ) . * 3 . * using the poincar inequality on each segment @xmath325 $ ] , and by ( [ contra ] ) : @xmath326 we now obtain convergence to @xmath68 of various quantities : @xmath327 consequently , by ( [ tre ] ) : @xmath328 hence : @xmath329 because of the equivalence of all norms on the finitely dimensional space @xmath78 , ( [ ml6 ] ) implies : @xmath330 now , we may estimate the quantity @xmath331 by the following norms : @xmath332 , @xmath333 , @xmath334 , @xmath335 , @xmath336 , @xmath337 , and use theorem [ approx ] , ( [ ml5 ] ) , ( [ sm1 ] ) , ( [ due.5 ] ) and ( [ ml7 ] ) to conclude that : @xmath338 together with ( [ ml2 ] ) this contradicts ( [ contra ] ) . in this section we prove theorem [ th_very_weak ] . the first step is to give a bound for the distance of @xmath10 from the generators of rigid motions in @xmath4 . this follows from theorem [ approx ] and the uniform poincar inequality ( [ th_uniform_poincare ] ) : [ cor4.2 ] assume * ( h1). for every @xmath211 there exists a linear function @xmath339 , @xmath128 , @xmath340 , such that : @xmath341 recall the results of theorem [ approx ] and define : @xmath342 by theorem [ approx ] and the poincar inequality on @xmath3 , we obtain : @xmath343 we now apply theorem [ th_uniform_poincare ] to the function @xmath344 , by which for some @xmath340 there holds : @xmath345 now ( [ 1.dieci ] ) and ( [ 1.undici ] ) imply the result . proof of theorem [ th_very_weak ] . * the proof of part ( i ) will be carried out assuming that @xmath105 on @xmath106 . for the other case ( @xmath105 on @xmath107 ) the argument is the same . we argue by contradiction . if ( [ very_weak ] ) was not true , then there would be sequences @xmath297 and @xmath298 satisfying the conditions in ( i ) or ( ii ) and such that : @xmath346 @xmath347 ( to simplify the notation , we write @xmath2 instead of @xmath299 ) . by corollary [ cor4.2 ] , there exists a sequence @xmath348 , @xmath349 , @xmath350 , such that : @xmath351 because of ( [ dodici ] ) , the sequence @xmath352 is bounded in @xmath23 and so , without loss of generality , we may assume that : @xmath353 moreover , by ( [ dodici ] ) and ( [ quattordici ] ) : @xmath354 and therefore : @xmath355 on @xmath106 , then we must have @xmath356 . indeed , by theorem [ th_uniform_trace ] and lemma [ trivial ] ( iv ) : @xmath357 where the convergence above follows from ( [ dodici ] ) and ( [ quattordici ] ) . thus : @xmath358 we now prove that if * ( h2 ) * holds , together with @xmath359 on @xmath19 , then @xmath360 . by theorem [ th_uniform_trace ] and lemma [ trivial ] ( v ) : @xmath361 where ( [ quattordici ] ) with ( [ 1.tredici ] ) justify the convergence . hence , by ( [ sedici ] ) : @xmath362 * 3 . * we see that in both cases ( i ) and ( ii ) there holds ( using condition ( [ dist_angle ] ) ) : @xmath363 thus , by ( [ quattordici ] ) and ( [ quindici ] ) : @xmath364 we deduce that @xmath365 as well , which contradicts ( [ ab_nonzero ] ) . in this section @xmath366 is a fixed open , bounded domain with lipschitz boundary . for @xmath367 , by @xmath368 we denote the outward unit normal to @xmath369 at @xmath218 . we first recall the standard korn inequality @xcite : [ korn_standard ] ( i ) there holds : @xmath370 and the following equivalence of norms : @xmath371 ( ii ) for every @xmath372 there exists @xmath128 and @xmath373 so that : @xmath374 the constants @xmath375 above depend only on the domain @xmath376 and not on @xmath10 . notice that theorem [ korn_standard ] ( ii ) implies that for each @xmath372 satisfying the orthogonality condition : @xmath377 one has : @xmath378 as we show below , the same is true if we restrict our attention to vector fields tangential on @xmath369 . define : @xmath379 [ naive_korn ] for every @xmath372 such that @xmath380 on @xmath369 and : @xmath381 there holds : @xmath382 and the constant @xmath375 depends only on @xmath376 . we argue by contradiction , starting with a sequence @xmath383 satisfying @xmath384 on @xmath369 , ( [ 6.2.uno ] ) and : @xmath385 without loss of generality , @xmath386 converges hence weakly to some @xmath10 in @xmath387 , and the convergence is strong in @xmath388 . clearly @xmath380 on @xmath369 and ( [ 6.2.uno ] ) still holds . by theorem [ korn_standard ] ( ii ) , there exist sequences @xmath389 and @xmath390 so that @xmath391 converges to @xmath68 in @xmath387 . therefore @xmath392 converges weakly to @xmath10 in @xmath387 and we see that @xmath393 . by ( [ 6.2.uno ] ) there hence must be @xmath18 and @xmath386 converges then ( strongly ) to @xmath68 in @xmath387 . this contradicts the first condition in ( [ help2 ] ) . let @xmath394 since @xmath395 , there must be @xmath396 , for some @xmath397 and we obtain : @xmath398 condition ( [ 6.2.uno ] ) reads : @xmath399 thus the class of functions @xmath10 for which the hypotheses of theorem [ naive_korn ] are satisfied is the following : @xmath400 as observed in the next result , condition ( [ 6.2.uno ] ) is not void if and only if our bounded domain @xmath376 is rotationally symmetric . [ lemma_rotations ] if @xmath401 then @xmath376 must be rotationally symmetric . let @xmath402 we will prove that the flow generated by the tangent vector field @xmath403 is a rotation . since @xmath128 we have that @xmath404 is an orthogonal decomposition of @xmath4 . write @xmath405 , @xmath406 , and consider the translated domain @xmath407 . now : @xmath408 so @xmath409 is a tangent vector field on @xmath410 . consider the flow @xmath118 which this field generates in @xmath4 : @xmath411 then @xmath412 , where : @xmath413 notice that : @xmath414 so @xmath415 remains bounded , while @xmath416 is unbounded for @xmath417 . since @xmath418 for all @xmath419 , there must be @xmath420 . hence the flow @xmath118 is a rotation ( generated by @xmath128 ) on @xmath410 , which proves the claim . from the proof above it follows that each @xmath421 has the form @xmath422 , @xmath128 , @xmath423 . we thus obtain the following characterisation when @xmath424 : @xmath425 throughout this section , @xmath376 is an open lipschitz domain in @xmath4 , star - shaped with respect to the origin and such that : @xmath426 for some @xmath427 and @xmath428 . the lipschitz constant of the boundary of @xmath376 is denoted by @xmath429 . our goal is to prove : [ app_main_korn ] for every @xmath372 there exists @xmath128 such that : @xmath430 and the constant @xmath431 above depends only on the quantities @xmath432 and @xmath429 . the proof is essentially a combination of the arguments in @xcite , where we need to keep track of the magnitude of various constants , and of @xcite . in @xcite , the @xmath24 distance of @xmath6 from a single proper rotation is estimated in terms of the @xmath24 norm of the pointwise distance of @xmath6 from the space of proper rotations @xmath433 . note that @xmath60 is the tangent space to @xmath433 at @xmath434 . hence ( [ app_main_korn ] ) can be seen as the `` linear '' version of the result in @xcite . [ app_kufner ] for every @xmath435 there holds : @xmath436 without loss of generality we may assume that @xmath437 . we adopt the proof of theorem 8.2 . in @xcite . let @xmath438 $ ] be a smooth non - decreasing function satisfying : @xmath439 fix a point @xmath440 and consider the function @xmath441 on the segment @xmath442 $ ] joining the origin and @xmath443 . using hardy s inequality and noting that @xmath444,\ ] ] we obtain : @xmath445 hence : @xmath446 which after integration in spherical coordinates gives : @xmath447 * 2 . * consider now a box @xmath448 such that @xmath449 and let @xmath450 $ ] be a smooth non - increasing function such that : @xmath451 we now apply hardy s inequality to the function @xmath452 on segments @xmath453 $ ] collinear with the @xmath454 direction , in the box @xmath455 inscribed in @xmath456 . taking @xmath457 , for all @xmath218 in the corresponding segment we have : @xmath458 using iterated integrals , we obtain : @xmath459 combining the inequalities ( [ app_kufner_1 ] ) and ( [ app_kufner_2 ] ) proves the lemma . [ app_fjm ] for every @xmath435 there exists @xmath460 such that : @xmath461 we adopt the method of proof from theorem 3.1 . in @xcite . for a small @xmath462 , consider the domain : @xmath463 by the poincar inequality on star - shaped domains we obtain : @xmath464 the above inequality follows , for example , from theorem [ th_chen_li ] in appendix d , in view of @xmath465 . therefore : @xmath466 applying lemma [ app_kufner ] to the function @xmath467 on @xmath376 we get : @xmath468 where the last inequality follows from ( [ app_fjm_2 ] ) . taking now @xmath469 sufficiently small , the first term in the right hand side is annihilated by the left hand side , which proves the lemma for @xmath470 . we now recall the following result from @xcite : [ app_oleinik ] let @xmath435 be such that @xmath471 in @xmath472 . then : @xmath473 * proof of theorem [ app_main_korn ] . * the left hand side of ( [ app_uno ] ) may be written as the distance ( in @xmath388 ) of @xmath6 from the closed subspace of constant functions @xmath128 . since the distance function is continuous , we may without loss of generality assume that @xmath474 . consider the problem : @xmath475 since : @xmath476 we see that : @xmath477 therefore : @xmath478 we are going to study the remaining part @xmath479 , which is harmonic : @xmath480 in @xmath376 . write : @xmath481 where each radius @xmath482 satisfies : @xmath483 in particular , this implies : @xmath484 it is also clear that the covering number of ( [ app_cov ] ) is bounded by some constant @xmath485 depending only on @xmath31 and @xmath429 . on each doubled ball @xmath486 we use the standard korn inequality ( see theorem [ korn_standard ] ) to obtain : @xmath487 for some matrices @xmath488 . use now lemma [ app_oleinik ] to the components of the harmonic vector function @xmath489 on @xmath486 : @xmath490 in view of ( [ app_proof_2 ] ) the above estimate implies : @xmath491 now by lemma [ app_fjm ] applied to the function @xmath183 , there exists @xmath492 so that : @xmath493 * 3 . * define @xmath494 and notice that for every @xmath495 there holds : @xmath496 therefore : @xmath497 now by ( [ app_proof_1 ] ) , ( [ app_proof_3 ] ) and ( [ app_proof_4 ] ) : @xmath498 the last inequality following from @xmath499 and the bound ( [ app_proof_1 ] ) . all the previous calculations were done assuming that @xmath500 . if @xmath501 , we may apply ( [ app_proof_5 ] ) to the rescaled domain @xmath502 and note that @xmath503 , while the lipschitz constant of @xmath504 remains the same as of @xmath369 . this proves the desired bound of the theorem . for a tangent vector field @xmath505 , define @xmath5 as the symmetric part of its tangential gradient : @xmath506.\ ] ] that is , for @xmath53 , @xmath507 is a symmetric bilinear form given through : @xmath508 recall that a smooth vector field @xmath10 as above is a killing field , provided that @xmath509 on @xmath3 . we first prove that in presence of this last condition , the regularity @xmath510 actually implies that @xmath10 is smooth . further , we directly recover a generalisation of theorem [ korn_standard ] ( ii ) to the non - flat setting ( theorem [ th_surface_korn ] ) . actually , the bound in theorem [ korn_standard ] ( i ) remains true also in the more general framework of riemannian manifolds @xcite . the following extension of @xmath10 on the neighbourhood of @xmath3 will be useful in the sequel : @xmath511 for some small @xmath512 . here @xmath513 is the shape operator on @xmath3 . we have @xmath514 where @xmath515 is open in @xmath4 . notice that for each @xmath516 and @xmath517 there holds : @xmath518\cdot ( \mbox{id } + t \pi(x))^{-1}\tau_1\big\}\cdot u(x)\\ & \qquad\qquad + ( \mbox{id } + t \pi(x))^{-1}\cdot\nabla u(x)\cdot ( \mbox{id } + t \pi(x))^{-1}\tau_1 . \end{split}\ ] ] the first component above is bounded by @xmath519 . taking the scalar product of the second component with any @xmath520 gives : @xmath521 since @xmath522 we obtain : @xmath523 on the other hand , @xmath524 , so for any @xmath83 : @xmath525 hence : @xmath526 [ weak_kil ] let @xmath505 be a tangent vector field such that @xmath21 almost everywhere on @xmath3 . then @xmath527 . we only need to prove that @xmath10 is smooth . consider the extension @xmath528 as above . by ( [ ext_d_1 ] ) , ( [ ext_d_2 ] ) and the formula ( [ useful_formula ] ) we see that @xmath529 and hence : @xmath530 the result follows now by the elliptic regularity and a bootstrap argument . [ th_surface_korn ] for every tangent vector field @xmath505 there exists @xmath116 such that : @xmath531 and the constant @xmath532 depends only on the surface @xmath3 . since @xmath78 is a finitely dimensional subspace of the banach space @xmath533 of all @xmath158 tangent vector fields , its orthogonal complement @xmath534 is a closed subspace of @xmath533 . we will prove that : @xmath535 which implies the theorem . if ( [ surf_uno ] ) was not true , there would be a sequence @xmath536 such that : @xmath537 without loss of generality , @xmath386 converge weakly in @xmath158 to some @xmath538 . moreover @xmath21 by the second condition above , so by lemma [ weak_kil ] we obtain that @xmath527 . as the spaces @xmath78 and @xmath534 are orthogonal , there must be @xmath18 , and hence the sequence @xmath386 converges to @xmath68 ( strongly ) in @xmath157 . this contradicts @xmath539 , because : @xmath540 the last inequality follows from theorem [ korn_standard ] ( i ) applied to the extensions @xmath541 as in ( [ def ] ) . indeed , by ( [ ext_d_1 ] ) and ( [ ext_d_2 ] ) it follows that : @xmath542 we now want to gather a few remarks relating to the fact that the linear space @xmath78 of all killing fields on @xmath3 is of finite dimension this is a classical result @xcite , and it implies that in @xmath78 all norms are equivalent . in particular , one has : @xmath543 for some constant @xmath532 depending only on the hypersurface @xmath3 . the bound ( [ s_uno ] ) , together with an estimate of @xmath532 , can also be recovered directly , using the following identity @xcite , valid for killing vector fields @xmath10 : @xmath544 here @xmath545 is the laplace - beltrami operator on @xmath3 , @xmath546 is the covariant derivative of @xmath10 on @xmath3 , and @xmath547 stands for the ricci curvature form on @xmath3 . to calculate @xmath548 in our particular setting , notice that by gauss teorema egregium ( @xcite , vol 3 ) , the riemann curvature @xmath549-tensor on @xmath3 satisfies : @xmath550 thus , seeing the ricci curvature @xmath551-tensor as an appropriate trace of @xmath164 , we obtain : @xmath552 integrating ( [ s_due ] ) on @xmath3 and using ( [ s_tre ] ) we arrive at : @xmath553 notice that in the special case of a @xmath554 matrix @xmath555 , that is when @xmath141 and @xmath3 is a 2-d surface in @xmath142 , the cayley - hamilton theorem implies : @xmath556 and so : @xmath557 in this case @xmath558 is the gaussian curvature of @xmath3 at @xmath218 ( see @xcite ) . to calculate the @xmath24 norm of the full gradient @xmath6 on @xmath3 , notice that : @xmath559 hence we arrive at : @xmath560 which clearly implies ( [ s_uno ] ) . an equivalent way of obtaining the formula ( [ s_cinque ] ) , but without using the language of riemannian geometry , is to look at the trivial extension of @xmath10 : @xmath561 since @xmath562 and @xmath563 on the boundary of @xmath515 , by ( [ useful_formula ] ) one has : @xmath564 calculating @xmath565 in terms of @xmath189 , dividing both sides of ( [ c_sei ] ) by @xmath566 and passing to the limit with @xmath567 , one may recover ( [ s_cinque ] ) directly . from the equivalence of the @xmath24 and the @xmath40 norms on @xmath78 , proven in ( [ s_cinque ] ) , it follows that the linear space @xmath568 is finitely dimensional . for otherwise the space @xmath569 would have a countable hilbertian ( orthonormal ) base @xmath570 and thus necessarily the sequence @xmath571 would converge to @xmath68 , weakly in @xmath158 . but this implies that @xmath572 , which by the norms equivalence gives the same convergence in @xmath158 , and a clear contradiction . [ th_chen_li ] assume that @xmath573 is open , lipschitz , star - shaped with respect to the origin and such that : @xmath574 then for every @xmath575 there holds : @xmath576 where the constant @xmath577 depends only on the quantities @xmath31 and @xmath578 . the proof follows , for example , from @xcite where the first nonzero eigenvalue @xmath579 of the neumann problem for @xmath580 on @xmath376 is estimated from below by @xmath581 , the constant @xmath582 depending on @xmath31 only . recalling that the best poincar constant equals to @xmath583 , the result follows . [ th_uniform_poincare ] assume * ( h1 ) * and let @xmath71 be sufficiently small . for every @xmath584 there exists a constant @xmath460 so that : @xmath585 and @xmath95 is independent of @xmath2 , @xmath586 or @xmath10 . first of all , by theorem [ th_chen_li ] , we see that the local estimate ( [ 1.uno ] ) is in our new setting replaced by : @xmath591 with @xmath592 and @xmath95 being , as usual , a uniform constant . repeating the calculations leading to ( [ 1.due ] ) and ( [ 1.tre ] ) , we thus obtain : @xmath593 which imply , exactly as in ( [ 1.cinque ] ) : @xmath594 now by the above inequalities and the standard poincar inequality on surfaces : @xmath595 which proves the result . [ th_uniform_trace ] assume * ( h1). for every @xmath596 there holds : @xmath597 @xmath598 where in the left hand side we have norms of traces of @xmath10 on @xmath3 and @xmath19 , respectively . the constant @xmath95 is independent of @xmath10 or @xmath2 . since @xmath599 , ( [ trace_uno ] ) will be implied by the same inequality for @xmath1 with @xmath600 . the latter one can be obtained covering @xmath1 with the cylinders @xmath215 of size @xmath2 and applying the scaled version of the usual trace theorem to @xmath215 . notice , that the constant @xmath95 in ( [ trace_uno ] ) depends only on @xmath31 and the lipschitz constant of @xmath3 . since @xmath601 and @xmath599 for each @xmath602 , we may use the same argument as before on @xmath603 to prove ( [ trace_due ] ) . a. korn , _ ber einige ungleichungen , welche in der theorie der elastischen und elektrischen schwingungen eine rolle spielen _ , bull . cracovie akademie umiejet , classe des sci . nat . , ( 1909 ) 705724 . solonnikov and v.e . scadilov , _ a certain boundary value problem for the stationary system of navier - stokes equations _ , boundary value problems of mathematical physics , 8 . trudy mat . inst . steklov . 125 * ( 1973 ) , 196210 , 235 .	we study the korn - poincar inequality : @xmath0 in domains @xmath1 that are shells of small thickness of order @xmath2 , around an arbitrary smooth and closed hypersurface @xmath3 in @xmath4 . by @xmath5 we denote the symmetric part of the gradient @xmath6 , and we assume the tangential boundary conditions : @xmath7 we prove that @xmath8 remains uniformly bounded as @xmath9 , for vector fields @xmath10 in any family of cones ( with angle @xmath11 , uniform in @xmath2 ) around the orthogonal complement of extensions of killing vector fields on @xmath3 . we show that this condition is optimal , as in turn every killing field admits a family of extensions @xmath12 , for which the ratio @xmath13 blows up as @xmath9 , even if the domains @xmath1 are not rotationally symmetric .
.[t : e2 ] electric quadrupole ( e2 ) clock transition . [ cols="^,^,^,^ " , ] another option is to use the ir@xmath3 ions . clock transitions in these ions involve hole states in the @xmath4 subshell which leads to extra enhancement of the sensitivity of clock frequencies to variation of the fine structure constant @xcite . a diagram for few lowest states of ir@xmath3 is presented on fig . [ f : ir17 ] . the energies and sensitivety coefficients for clock states of ir@xmath3 ions are presented in table [ t : ir17 ] . note , that the ion has two clock transitions which have different sensitivity to variation of @xmath8 . if the ratio of two frequencies is measured , the combined sensitivity is very large @xmath15 experimental work with ir@xmath3 ions is in progress at max planck institute @xcite . it has been mentioned above that we are looking for systems with large sensitivity coefficients @xmath16 and clock frequency @xmath17 beeing in optical region so that the enhancement factor @xmath18 ( @xmath19 ) is large due to large @xmath16 rather than small @xmath17 . the largest sensitivity factors found so far are in the cf@xmath3 and cf@xmath7 ions @xcite . however , both these ions are not very convenient for the use as atomic clocks . the cf@xmath3 ion has only one optical transition other than clock transition . this is a magnetic dipole transition between ground @xmath20 state and excited @xmath21 state . it is weak due to small frequency of the transition . this makes it difficult working with the ion . the cf@xmath7 ion has only one metastable excited state which can serve as a clock state . the transition between ground @xmath22 and metastable @xmath23 states is the magnetic octupole ( m3 ) transition . it is so weak that its use as clock transition is very problematic . it turns out that good clock transitions exist in the cf@xmath6 ion as well as in the es@xmath7 and es@xmath3 ions . the main parameters of the ions and corresponding clock transitions are summarized in table [ t : cfes ] . note that both elements have only unstable isotopes . however , many isotopes have very long lifetime ( see , e.g. table [ t : cfes ] ) . the choice of isotopes in table [ t : cfes ] is dictated by two considerations . first , it is a long - living isotope . second , nuclear spin @xmath24 has such a value that it is always possible to have @xmath25 , @xmath26 for both states in the clock transition . here @xmath27 is the total angular momentiom of the ion and @xmath28 is its projection . states with @xmath25 and @xmath26 have zero quadrupole moment and are not sensitive to gradients of electric field . the authors are gratefull to h. katori and m. wada for useful discussions . the work was supported by the australian research council . v. a. dzuba , a. derevianko , and v. v. flambaum , phys . a * 86 * , 054502 ( 2012 ) . m. s. safronova , v. a. dzuba , v. v. flambaum , u. i. safronova , s. g. porsev , and m. g. kozlov , phys . 113 * , 030801 ( 2014 ) . a. windberger , o. o. versolato , h. bekker , n. s. oreshkina , j. c. berengut , v. bock , a. borschevsky , v. a. dzuba , z. harman , s. kaul , u. i. safronova , v. v. flambaum , c. h. keitel , p. o. schmidt , j. ullrich , and j. r. crespo lopez - urrutia , identification of optical transitions in complex highly charfed ions for applications in metrology and tests of fundamental constants . to be published .	we review a number of highly charged ions which have optical transitions suitable for building extremely accurate atomic clocks . this includes ions from hf@xmath0 to u@xmath1 , which have the @xmath2 configuration of valence electrons , the ir@xmath3 ion , which has a hole in almost filled @xmath4 subshell , the ho@xmath5 , cf@xmath6 , es@xmath3 and es@xmath7 ions . clock transitions in most of these ions are sensitive to variation of the fine structure constant , @xmath8 ( @xmath9 ) . e.g. , californium and einsteinium ions have largest known sensitivity to @xmath8-variation while holmium ion looks as the most suitable ion for experimental study . we study the spectra of the ions and their features relevant to the use as frequency standards . highly charged ions ( hci ) can be used for building a new generation of very accurate optical clocks @xcite . this may have many technical applications but also clock transitions in hci can be used to study fundamental problems of modern physics such as variation of fundamental constants @xcite , local lorentz invariance violation @xcite , search for dark matter @xcite , etc . having extremely high accuracy of the clocks is crucial for these studies . clocks quality factor can be defined as a ratio of clock frequency to the value of its perturbations @xmath10 currenf microwave cesium clock , which serves as definition of metric second , have @xmath11 @xcite , best optical clocks approach @xmath12 @xcite mostly due to larger frequency . further progress can be achieved by using optical transitions in hci @xcite . hci are less sensitive to perturbations due to thier compact size . therefore , quality factor @xmath13 can be larger than in neutral atoms due to smaller @xmath14 . in this paper we review some recent proposals for very accurate atomic clocks based on hci and thier use for the search of time variation of the fine structure constant .
the steepest descent method for asymptotic analysis of matrix riemann - hilbert problems was introduced by deift and zhou in 1993 @xcite . a matrix riemann - hilbert problem is specified by giving a triple @xmath1 consisting of an oriented contour @xmath2 in the complex @xmath3-plane , a matrix function @xmath4 which is usually taken to be continuous except at self - intersection points of @xmath2 where a certain compatibility condition is required , and a normalization condition @xmath5 as @xmath6 . if @xmath2 is not bounded , certain asymptotic conditions are required of @xmath7 in order to have compatibility with the normalization condition . consider an analytic function @xmath8 taking continuous boundary values @xmath9 ( respectively @xmath10 ) on @xmath2 from the left ( respectively right ) . the riemann - hilbert problem @xmath1 is then to find such a matrix @xmath11 satisfying the normalization condition @xmath5 as @xmath6 and the jump condition @xmath12 whenever @xmath3 is a non - self - intersection point of @xmath2 ( so the left and right boundary values are indeed well - defined ) . the steepest descent method of deift and zhou applies to certain riemann - hilbert problems where the jump matrix @xmath13 depends on an auxiliary control parameter , and is a method for extracting asymptotic properties of the solution @xmath11 ( and indeed proving the existence and uniqueness of solutions along the way ) when the control parameter tends to a singular limit of interest . the original method put forth in @xcite bears a striking resemblance to the well - known steepest descent method or saddle point method for analyzing contour integrals with exponential integrands . a distinguished point on @xmath2 is identified ( analogous to a point of stationary phase ) and an explicit change of variables of the form @xmath14 where @xmath15 is a piecewise analytic matrix is introduced in the vicinity of this point and it is observed that ( i ) the matrix @xmath16 satisfies an equivalent riemann - hilbert problem with a new contour @xmath17 and a new jump matrix @xmath18 , and ( ii ) the jump matrix @xmath19 converges to the identity matrix in the singular limit of interest for all @xmath3 bounded away from the stationary phase point . one therefore expects that a good approximation to @xmath20 can be constructed by an explicit local analysis near the stationary phase point . with the explicit local approximant @xmath21 constructed , one uses it in a final change of variables @xmath22 and observes ( i ) that @xmath23 satisfies a riemann - hilbert problem with a possibly new contour @xmath24 and a new explicit jump matrix @xmath25 and ( ii ) that the new jump matrix @xmath26 is now _ uniformly _ close to the identity matrix in the limit of interest . this allows one to construct @xmath23 by iteration of certain singular integral equations that are equivalent to any given riemann - hilbert problem , and to show that @xmath23 is uniformly close to the identity matrix in any region bounded away from @xmath24 . with additional work it may in some circumstances be shown that @xmath23 is close to the identity uniformly right up to the contour @xmath24 ( this usually requires more detailed information about the jump matrix ) . in this way , one obtains a formula @xmath27 for the solution that can be used to compute _ directly _ an asymptotic expansion of @xmath28 valid in the singular limit of interest . since the introduction of the steepest descent method for riemann - hilbert problems , there have been several key developments . in @xcite and @xcite a technique was established in which one makes a change of variables involving a matrix constructed from a single unknown scalar function @xmath29 analytic in @xmath30 . the transformation modifies the jump matrix in a way involving the boundary values @xmath31 taken on @xmath2 . one then chooses relations between the boundary values of @xmath29 such that the transformed riemann - hilbert problem becomes asymptotically simple . the desired conditions amount to a _ scalar _ riemann - hilbert problem for @xmath29 , which is easily solved in many circumstances . a crucial feature of this method is that the dominant contribution to the solution typically comes from subintervals of the contour @xmath2 of finite length rather than from isolated points . here we therefore see an important difference between singular limits of matrix riemann - hilbert problems and evaluation of saddle - point integrals . in the contributing intervals the transformed jump matrix has a factorization ( see , , and below ) whose factors admit analytic continuation to the left and right of each such interval . a further change of variables based on this analytic factorization is carried out in lens - shaped regions surrounding each contributing interval . ultimately , a model problem is solved ( typically in terms of riemann theta functions of genus related to the number of contributing intervals ) and along with local analysis near the endpoints of the intervals a model for @xmath11 is built and compared with @xmath11 to obtain a riemann - hilbert problem for the error @xmath23 . when the method is successful , the jump matrix for @xmath23 is uniformly close to the identity and thus @xmath32 may be constructed via iteration of integral equations . significantly , the conditions imposed on the boundary values of @xmath29 can often be viewed as the euler - lagrange conditions for a certain variational problem ( see @xcite as well as @xcite and appendix [ app : potential ] of this paper ) . a further development emerged from problems in which it was recognized that no appropriate function @xmath29 can be found relative to the given contour @xmath2 . in @xcite and later in @xcite it was shown how analyticity of the jump matrix could be exploited to effectively deform arcs of the contour @xmath2 to alternative locations in the complex plane such that the jump matrix maintains the same functional form ; specific locations of the arcs are determined such that there exists an appropriate function @xmath29 as above . these selected arcs are the closest relatives in the noncommutative theory to the paths of steepest descent from saddle points in the asymptotic theory of contour integrals . the contour selection principle was also encoded into a variational problem in @xcite . more recently @xcite , new techniques have been added to the framework of the steepest descent method that are adapted for determining the asymptotic contribution to the solution of a coalescence of a large number of poles in the matrix unknown ( this is strictly speaking not a riemann - hilbert problem in the sense described above due to the polar singularities , however the problem is first converted into a standard riemann - hilbert problem by explicit transformations ) . the key idea here is to exploit certain analytic interpolants of given residues at the poles . for the fundamentally nonlinear cases in which the dominant contribution comes from subintervals of a contour , a central feature is that the analytical methods rely on piecewise analyticity of the given jump matrix and of the boundary values of the scalar function @xmath29 . for some of the cases of long - time asymptotics of integrable nonlinear partial differential equations @xcite , as well as the recent long time asymptotic analysis for perturbations of the defocusing nonlinear schrdinger equation @xcite , the dominant contribution comes from isolated points of the contour @xmath2 , and while analyticity is not fundamental , the asymptotic calculations proceed by an approximation argument , in which an analytic part is deformed away , and a ( small ) residual contribution is handled by technical and analytical prowess . the approximation argument is delicate and requires detailed analysis that depends sensitively on the geometry of the particular contour @xmath2 . far from being a mere pursuit of abstraction , a simple asymptotic technique that applies to riemann - hilbert problems regardless of whether the jump matrix is analytic or not would have immediate application in a number of important areas . for example , a unified treatment of the asymptotic theory of orthogonal polynomials on the real line with general nonanalytic weights would allow the resolution of universality conjectures from random matrix theory in the most natural and general context ( see @xcite for the analytic case , and @xcite for an application of the @xmath0 steepest descent method described in this paper to the nonanalytic but convex case ) . as another example , if it were possible to treat systematically problems with a large number of poles that accumulate in a very regular but nonanalytic fashion , then the important problem of semiclassical asymptotics for the focusing nonlinear schrdinger equation with general nonanalytic initial data could begin to be addressed . in this paper , we present a new generalization of the steepest descent method for riemann - hilbert problems that applies in absence of analyticity of the jump matrix , and yet does not depend on an approximation argument for the jump matrix . while we believe that the ideas we will develop in this paper are useful in very general contexts , we have chosen to focus on a particular application of the steepest descent method in order to demonstrate the technique . as mentioned above , after changing variables using an appropriate scalar function @xmath29 , the jump matrix is converted into a form that is well - suited for further asymptotic analysis . a common `` target '' form for the jump matrix in certain arcs of @xmath2 is the following form @xmath33 where @xmath34 is a real parameter along the arc of @xmath2 which for simplicity here we assume lies on the real axis ( for a representation see figure [ fig : intro ] ) , @xmath35 is the control parameter tending to zero in the singular limit of interest , and @xmath36 is a strictly increasing real function of @xmath34 that is related to the boundary values of @xmath29 on @xmath2 . suppose @xmath16 is the unknown satisfying @xmath37 . with the assumption of analyticity of @xmath36 , one may transform the riemann - hilbert problem by introducing as a new unknown a matrix @xmath38 defined in terms of @xmath16 by the following scheme : in some region lying on the minus side of the arc ( the region labeled @xmath39 in figure [ fig : intro ] ) , set @xmath40 and in some region lying on the plus side of the arc ( the region labeled @xmath41 in figure [ fig : intro ] ) , set @xmath42 elsewhere , set @xmath43 . on has thus introduced two new jump contours , one on either side of the arc ( these are the two dashed lines in figure [ fig : intro ] ) . however , the monotonicity of the real analytic function @xmath36 implies via the cauchy - riemann equations that the induced jump matrix relating the boundary values of @xmath44 on these two contours is exponentially close to the identity matrix in the limit @xmath45 . on the original arc , the matrix @xmath38 satisfies the constant jump relation @xmath46 which can be subsequently dealt with in terms of special functions . in this paper , we show how this procedure can be carried out effectively when one does not have the option of extending @xmath36 from the contour @xmath2 because it is not assumed to be an analytic function . we choose to extend @xmath36 in a way that does not assume any analyticity ( see below ) . the price that must be paid is that the analogue of the matrix @xmath44 above is no longer analytic in the regions to the left and right of the arc ; therefore this matrix can not be the solution of any riemann - hilbert problem . it can , however , be the solution of a matrix @xmath0 problem ( or more generally a mixed riemann - hilbert-@xmath0 problem ) . it is into this framework that we extend the steepest descent method . this explains the terminology of the `` @xmath0 steepest descent method '' . as the fundamental contour @xmath2 in this paper is the unit circle @xmath47 , we can now be very specific about what we mean by an extension of a nonanalytic function in this context . suppose that @xmath48 is a @xmath49 function , @xmath50 . then , we define an extension operator @xmath51 as follows : @xmath52 where @xmath53 are the standard polar coordinates for @xmath54 . note that this indeed defines a continuous extension to any annulus @xmath55 where @xmath56 since @xmath57 . also , since @xmath58 and @xmath59 , the fundamental differential operators of complex variable theory are represented in polar coordinates as @xmath60 and therefore we see that if @xmath61 is lipschitz , then in particular it has a derivative almost everywhere that is uniformly bounded , and we have @xmath62 and @xmath63 both holding for all @xmath64 and almost all @xmath65 ( these formulae hold at every point of the plane if @xmath66 is of class @xmath67 ) . it follows that @xmath68 vanishes to order @xmath69 as @xmath70 uniformly in @xmath71 . in fact , if @xmath48 is analytic for all @xmath71 , then the infinite series @xmath72 converges uniformly in some annulus containing the unit circle @xmath73 and represents the unique analytic extension of @xmath48 . generally speaking , riemann - hilbert problems with rapidly oscillatory jump matrices are equivalent to systems of singular integral equations with cauchy kernel and rapidly oscillatory densities . such equations can in principle be analyzed asymptotically @xcite . this approach requires delicate arguments of harmonic analysis . on the other hand , the @xmath0 steepest descent method we will develop in this paper avoids such complicated reasoning . indeed , by extending contour integration into integration over two - dimensional regions , the cauchy kernel becomes less singular , and the analysis becomes correspondingly more straightforward . in the analytic case described briefly above , the asymptotic analysis is in general complicated by the fact that the procedure is valid in the neighborhood of certain intervals of @xmath2 , and it turns out that a different analysis must be carried out in the vicinity of the endpoints of the intervals . the same would be expected to be true in the general nonanalytic case . in order to have the clearest possible presentation , we have chosen to describe in this paper the @xmath0 steepest descent method in the context of a problem where there are nontrivial cases without endpoint issues , namely the asymptotic behavior of polynomials orthogonal with respect to weights on the unit circle . while the presence of endpoints complicates the analysis , they do not present an insurmountable obstruction , and the reader is referred to @xcite for a description of the more general theory . let @xmath74 be an integrable @xmath75-periodic function satisfying @xmath76 for almost all @xmath71 . for two complex - valued functions @xmath48 and @xmath77 there is an associated inner product @xmath78 this inner product leads to a system @xmath79 of orthogonal polynomials in the complex variable @xmath3 : @xmath80 and the defining relation is the orthonormality condition @xmath81 here the polynomials @xmath82 are considered as complex - valued functions of @xmath71 by restriction to the unit circle : @xmath83 . the constants @xmath84 have the interpretation of normalization constants , and the corresponding system @xmath85 of monic orthogonal polynomials is defined by rescaling : @xmath86 the orthogonal polynomials satisfy recurrence relations of the form @xmath87 for @xmath88 . here , the complex constants @xmath89 are the _ recurrence coefficients _ associated with the weight @xmath90 ( also known as the _ schur parameters _ or _ verblunsky coefficients _ ) . by setting @xmath91 in the first equation of , it is easy to see that @xmath92 for a general discussion of properties of polynomials orthogonal on the unit circle , see chapter xii of szeg s monograph @xcite , in which ( among other things ) the asymptotic behavior of @xmath93 for @xmath94 is discussed . the extraction of asymptotic formulae for quantities related to the orthogonal polynomials of large degree with a fixed weight @xmath90 on the unit circle is the type of asymptotic problem in the theory of general orthogonal polynomials ( that is , beyond particular cases involving classical special functions ) for which results have been known for the longest time . this can be traced to the fact that if @xmath95 happens to be a positive trigonometric polynomial , then there is a closed - form expression for the orthonormal polynomial @xmath82 that is convenient for analysis , as long as @xmath96 is sufficiently large compared to the degree of @xmath95 . see @xcite . in other words , for certain special fixed weights @xmath74 the asymptotic formulae one obtains become _ exact _ as long as @xmath96 is large enough . this leads to a general strategy for asymptotic analysis of orthogonal polynomials on the unit circle based on approximating an arbitrary given positive function @xmath95 by positive trigonometric polynomials . the asymptotics described in the monograph of szeg are of a rather general character and hold whenever @xmath97 is an integrable real valued function . in the years since the origin of szeg s methods , there have been many further developments in the asymptotic theory . these developments move both in the direction of generalizing the class of weights for which the szeg asymptotics are valid ( perhaps in a weaker form ) and also in the direction of trading generality of the weight for detail of the asymptotics . it seems that certain problems remain difficult to treat by these methods ; in particular , it is difficult to verify convergence in a uniform sense , and it is difficult to characterize the detailed asymptotic behavior of zeros . there is a vast literature on this subject ; we refer the interested reader to the memoir of nevai @xcite and the forthcoming monograph of simon @xcite . beyond being a source of classical information about orthogonal polynomials on the unit circle , simon s monograph describes a different viewpoint of the theory of these polynomials . namely , simon and his school have made great progress by exploiting the connection between orthogonal polynomials and spectral theory for operators that encode the recurrence relations that all orthogonal polynomials satisfy ( see also @xcite ) . this theory is capable of establishing a number of very general results relevant to asymptotics in the limit of large degree . an important point is that the hypotheses required to establish results of this kind involve assumptions about the asymptotic behavior of the sequence @xmath89 of recurrence coefficients . indeed , the fundamental problem of spectral theory in this context is the construction of the spectral measure @xmath98 from the finite difference operator involving the recurrence coefficients @xmath89 . on the other hand , the recovery of the polynomials @xmath79 and of the recurrence coefficients @xmath89 from the spectral measure @xmath98 is the fundamental problem of inverse spectral theory . a general approach to asymptotic problems in the theory of orthogonal polynomials in which the measure of orthogonality is the given data therefore involves the translation of the orthogonality conditions into the conditions making up a riemann - hilbert problem for sectionally analytic matrices . a riemann - hilbert formulation for polynomials orthogonal with respect to a measure on the unit circle was described in @xcite and follows closely the well - known riemann - hilbert formulation for polynomials orthogonal with respect to a measure on @xmath99 discovered in @xcite . in @xcite , and in a number of papers which followed ( see , for example , @xcite ) , the polynomial of degree @xmath96 orthogonal with respect to a specific family of weights of the form @xmath100 where @xmath101 was studied in the limit @xmath102 . note that this is a joint limit as the degree @xmath96 of the polynomial in question appears in the measure of orthogonality as well ; see [ sec : varying ] for a general discussion of such exponentially varying weights . for the large @xmath96 asymptotics carried out in @xcite , and in subsequent works with this measure , as well as closely related measures ( see , for example , @xcite ) , analyticity of the weight @xmath74 played a central role in the analysis . in @xcite , deift used polynomials orthogonal with respect to a measure on the unit circle to give an example of his theory of integrable operators . specifically , he introduced a one - parameter family of positive , analytic functions @xmath103 and related solutions of riemann - hilbert problem [ rhp : m ] in [ sec : ops ] below ( with @xmath74 replaced by @xmath103 ) to toeplitz determinants . the @xmath94 asymptotic behavior of the corresponding solution @xmath104 to riemann - hilbert problem [ rhp : m ] then yields asymptotics for the associated toeplitz determinants . exploiting the analyticity of @xmath103 , deift outlined how one obtains an asymptotic description for @xmath105 . the calculations in we will present in [ sec : strong ] may be viewed as complementary to this asymptotic calculation of @xcite , in that we will establish asymptotics for orthogonal polynomials under the much weaker assumption that @xmath74 is a continuous function satisfying a lipschitz condition . furthermore , we show how the error estimates depend on smoothness properties of @xmath74 . the polynomials orthogonal with respect to a weight given on the unit circle in the complex plane can be characterized in terms of the solution of a matrix riemann - hilbert problem in which the contour @xmath2 is the unit circle . in [ sec : ops ] we describe this riemann - hilbert problem , and then in [ sec : strong ] and [ sec : varying ] we study the singular limit in which the degree of the polynomials tends to infinity . in [ sec : strong ] we consider the weight function to be held fixed as the degree tends to infinity , while in [ sec : varying ] we study the joint limit when the degree becomes large while the weight function is exponentially varied . a summary of the relevant logarithmic potential theory referred to in [ sec : varying ] is given in appendices [ app : potential ] and [ app : compare ] . the key results we obtain in the fixed weights case are described in [ sec : strongasymptoticspi ] . to the best of our knowledge , the uniform nature of the asymptotics we obtain is new to the field , as is our detailed characterization of the zeros . while there exist several classical methods available for the asymptotic analysis of orthogonal polynomials on the unit circle with fixed weight function @xmath74 , with the degree @xmath96 of the polynomial in question tending to infinity , this problem is the ideal context in which to introduce the @xmath0 steepest descent method . on the other hand , the asymptotic behavior of polynomials orthogonal with respect to a varying weight on the unit circle , considered in [ sec : varying ] , is more challenging to obtain by more classical techniques . the results we obtain with the use of the @xmath0 steepest descent method are stated in [ sec : asymptoticspids ] . a point we wish to emphasize is that with the use of the @xmath0 steepest descent method , the analysis in the varying weights case is no more difficult than in the fixed weights case . this fact distinguishes the @xmath0 steepest descent method from more classical techniques . throughout , we assume a fixed norm @xmath106 on @xmath107 matrices . for @xmath108 , we use the following induced norm on sufficiently smooth matrix functions @xmath109 @xmath110 here @xmath34 and @xmath111 are cartesian coordinates in @xmath54 . for @xmath112 , we may apply this norm on all functions @xmath113 in the space @xmath114 , whereas for @xmath115 we may apply this norm to a subset of functions @xmath113 in the space @xmath116 of functions with lipschitz continuous mixed partial derivatives of all orders up to and including @xmath117 . the finiteness of the norm indicates the uniformity of the lipschitz condition . since for functions @xmath113 in the class @xmath116 the mixed partial derivatives of order @xmath118 exist almost everywhere , the condition @xmath119 can be equivalently expressed as saying that @xmath113 has all derivatives of total order at most @xmath118 in the space @xmath114 . we also use the notation @xmath120 and @xmath121 for spaces of functions respectively in @xmath116 and @xmath114 that vanish identically for @xmath122 large enough ( that is , outside some annulus ) . for functions @xmath123 defined on the circle @xmath47 ( that is , @xmath124 is defined for @xmath125 ) , we also say that @xmath124 is of class @xmath126 if the periodic extension of @xmath124 to @xmath127 has @xmath128 lipschitz continuous derivatives , or equivalently , has @xmath129 derivatives in @xmath130 . a suitable norm for such functions is given by @xmath131 satisfying @xmath132 , @xmath133 if @xmath123 satisfies a hlder continuity condition , there exists a unique function @xmath134 , analytic for @xmath135 , decaying as @xmath6 , and taking hlder continuous boundary values on @xmath136 , such that @xmath137 and @xmath138 is the average value of @xmath124 . thus , @xmath139 is the negative frequency component of the fourier series for @xmath123 : @xmath140 and we have @xmath141 we also introduce the function @xmath142 by the formula @xmath143 note that @xmath144 and @xmath124 are functions that are related by the cauchy transform . throughout the paper we will use a `` bump '' function @xmath145 $ ] with the properties that @xmath146 is infinitely differentiable , @xmath147 for @xmath148 , and @xmath149 for @xmath150 . consider the contour @xmath2 illustrated in figure [ fig : circle ] . let @xmath96 be a positive integer . relative to the contour @xmath2 we pose , for each @xmath88 , the following riemann - hilbert problem for a @xmath107 matrix @xmath151 . find a @xmath107 matrix @xmath152 with the properties : * * analyticity . * @xmath152 is analytic for @xmath153 , and takes continuous boundary values @xmath154 , @xmath155 as @xmath156 tends to @xmath3 with @xmath136 and @xmath157 , @xmath158 . * * jump condition . * the boundary values are connected by the relation @xmath159 * * normalization . * the matrix @xmath152 is normalized at @xmath160 as follows : @xmath161 [ rhp : m ] suppose that the positive weight function @xmath74 satisfies a uniform hlder condition @xmath162 for some @xmath163 $ ] and with some @xmath164 independent of @xmath165 and @xmath166 . then riemann - hilbert problem [ rhp : m ] has a unique solution for each integer @xmath167 , namely if @xmath168 , @xmath169 and if @xmath170 , @xmath171 in particular , @xmath172 and @xmath173 for @xmath174 . here , @xmath175 is the sequence of monic orthogonal polynomials with respect to the weight @xmath90 and the inner product , and @xmath89 is the sequence of associated recurrence coefficients ( see ) while @xmath176 is the sequence of associated normalization constants . [ prop : m ] if @xmath170 , then the riemann - hilbert problem is triangular with identity asymptotics and is trivially solved in closed form by a cauchy integral , yielding . thus from now on we consider @xmath174 . the uniqueness of the solution for @xmath174 can be seen from the following argument . continuity of the boundary values taken on @xmath2 implies that the ratio of any two solutions of riemann - hilbert problem [ rhp : m ] is an entire function of @xmath3 that tends to the identity matrix as @xmath6 . uniqueness thus follows by liouville s theorem . to derive , first note that if @xmath151 solves riemann - hilbert problem [ rhp : m ] , then the first column of @xmath152 must be analytic throughout the @xmath3-plane . from the normalization condition ( [ eq : mnorm ] ) it is then clear that @xmath177 is a monic polynomial of degree @xmath96 while @xmath178 is a polymomial of degree at most @xmath179 ( the leading coefficient of @xmath178 is not determined from the normalization condition alone ) . the jump condition for the second column reads @xmath180 in particular , since @xmath74 satisfies a hlder condition , we may express @xmath181 as a cauchy - type integral using ( [ eq : seccoljump ] ) . thus , @xmath182 and the normalization condition ( [ eq : mnorm ] ) then requires that this cauchy integral be of order @xmath183 as @xmath6 for each fixed @xmath184 . expanding the cauchy kernel in a geometric series we see that the following conditions must be satisfied @xmath185 for @xmath186 . since @xmath187 for an angular coordinate @xmath71 on the contour @xmath188 , this proves that @xmath189 is orthogonal to the monomials @xmath190 , @xmath3 , @xmath191 , @xmath192 , @xmath193 with respect to the inner product . the existence of such a monic polynomial of degree @xmath96 follows from the gram - schmidt algorithm . thus , @xmath194 , the @xmath96th monic orthogonal polynomial with respect to the weight @xmath74 on the unit circle . a similar argument applies to the second row of @xmath152 . indeed , @xmath178 is a polynomial of degree at most @xmath179 . using ( [ eq : seccoljump ] ) we may express @xmath195 as a cauchy integral : @xmath196 and then the normalization condition ( [ eq : mnorm ] ) requires that @xmath197 as @xmath6 . expanding the cauchy kernel in a geometric series , one sees that @xmath178 is required to satisfy the following conditions : @xmath198 equivalently , these relations may be written in the form @xmath199 clearly , the degree @xmath179 polynomial @xmath200 is orthogonal to the monomials @xmath201 with respect to the inner product @xmath202 , and the normalization condition then fixes the leading coefficient : @xmath203 . in other words , we have found that @xmath204 , or equivalently , @xmath205 . thus , the conditions of riemann - hilbert problem [ rhp : m ] serve to define the orthogonal polynomials as an alternative to other representations that may be available , possibly including explicit contour integral formulae ( for special families of weights ) . riemann - hilbert problems like this one frequently arise as a consequence of the application of the fourier transform or @xmath3-transform to certain types of linear integral equations ( _ i.e. _ the wiener - hopf technique , see @xcite ) . reversing this sort of reasoning , the representation of the orthogonal polynomials in terms of riemann - hilbert problem [ rhp : m ] immediately yields integral equations for certain auxiliary unknowns . of particular interest are the marchenko equations obtained in @xcite , in which the unknowns are fourier coefficients of functions explicitly related to the orthogonal polynomial @xmath206 , and the relevant operator is of the form @xmath207 , where @xmath164 is an integral operator acting in @xmath208 with a kernel that depends explicitly on the weight @xmath90 but not otherwise on the degree @xmath96 of the polynomial in question . furthermore , the kernel does not depend in any crucial way on the smoothness of the weight . therefore , in principle , this formulation makes possible the calculation of asymptotics for the polynomials ( indeed , this is one of the applications of the marchenko equations discussed in @xcite ) in a way that is relatively insensitive to the analyticity properties of the weight function @xmath90 . however , the correction terms that appear in such a scheme are necessarily in terms of infinite fourier series ( see , _ e.g. _ , equation ( vi.7 ) of @xcite ) that while having known coefficients are not convenient for detailed analysis of zeros of the polynomials in regions of the complex plane where these zeros necessarily arise from a competition between different terms in an expansion . to provide details of the asymptotics , it is more advantageous to work with the riemann - hilbert problem directly . the uniqueness of @xmath152 coupled with symmetry of the riemann - hilbert problem under reflection through the unit circle leads to the following result . the matrix @xmath152 satisfying riemann - hilbert problem [ rhp : m ] satisfies the symmetry relation @xmath209 in particular , by taking @xmath91 above , we obtain the identity @xmath210 [ prop : gammaalpha ] in this section we describe several asymptotic results in the theory of orthogonal polynomials with fixed weights on the unit circle that we will obtain as a fundamental illustration of our method . it will be convenient to introduce the real - valued function @xmath211 such that @xmath212 the fundamental object of the asymptotic theory for the fixed weight @xmath90 is the so - called szeg function : @xmath213 this is a function analytic for @xmath153 that decays to zero as @xmath6 . its value at @xmath91 has the interpretation of the geometric mean of the weight @xmath74 : @xmath214 if @xmath74 satisfies a hlder continuity condition , then by strict positivity so does @xmath123 . in this case , by the plemelj formula @xcite , we have @xmath215 furthermore , recalling the negative frequency component @xmath134 of @xmath124 defined by , we can obtain the following equivalent characterization of @xmath216 . since the function @xmath217 is analytic for @xmath218 , it follows that @xmath219 recalling the function @xmath142 defined by , we have @xmath220 the following results hold for weights where @xmath124 is of class @xmath126 with @xmath221 . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath224 . then , for each fixed integer @xmath118 and for each @xmath225 there is a constant @xmath226 such that the estimate @xmath227\right\| \le k_{p,\rho}\frac{\log(n)}{n^{2k } } \label{eq : piasympoutside}\ ] ] holds for all @xmath96 sufficiently large . [ thm : pioutside ] the constant @xmath228 typically blows up as @xmath229 , and only a finite number of derivatives can be controlled . more generally , we have the following result . let @xmath230 be a fixed integer . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath231 . then there exists a constant @xmath232 such that the estimate @xmath233\right\|\le k_p\frac{\log(n)}{n^{k-2p } } \label{eq : pialloutsideestimate}\ ] ] holds for all @xmath96 sufficiently large . [ thm : picircle ] note that as a special case of the estimate we obtain the following estimate ( under the same conditions ) characterizing the polynomials on the unit circle : @xmath234\right\|\le k_p\frac{\log(n)}{n^{k-2p}}\ , . \label{eq : picircleestimate}\ ] ] in fact , the proof of theorem [ thm : picircle ] is to first establish , from which the estimate follows ( with the same constant @xmath235 ) via the maximum modulus principle . the weakest conditions under which the above theorem provides large - degree asymptotics are that @xmath90 is a strictly positive weight that is lipschitz continuous . theorem [ thm : picircle ] may be compared with results reported in the classic monograph of szeg @xcite . while asymptotics of @xmath206 have been established by other methods under weaker conditions than lipschitz continuity and strict positivity of the weight @xmath90 , theorem [ thm : picircle ] exhibits clearly the dependence of the rate of decay of the error on the smoothness of @xmath90 , and the number of derivatives desired . to our knowledge , the results of theorem [ thm : picircle ] are stronger than those previously known in that they establish the convergence in a uniform sense . this leads to the following . let @xmath230 be a fixed integer . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath236 . then @xmath237 this follows directly from theorem [ thm : picircle ] . indeed , upon carrying out the differentiation in , and combining this estimate with its analogues for all smaller values of @xmath118 , we learn that @xmath238 converges uniformly to @xmath239 for @xmath240 . the proof is then complete since on @xmath47 uniform convergence implies convergence in @xmath241 . notice that theorem [ thm : picircle ] immediately implies the following formula valid for all @xmath3 with @xmath242 : @xmath243 where @xmath244 while in principle this could be used to compute asymptotics for @xmath93 in this region , more detailed analysis gives the following improved results . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath221 . then for each @xmath245 satisfying @xmath246 there are constants @xmath247 such that the estimates @xmath248 and @xmath249 hold for all @xmath96 sufficiently large . [ thm : piinside ] an immediate corollary is that there exists an annulus inside the unit circle that asymptotically contains no zeros . that the result we are about to state in this direction is in a sense sharp will be made clear when we consider more specific weights below in [ sec : jumpweights ] ( in particular , see corollary [ cor : jumpzeroscircle ] ) . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath221 . let @xmath250 be an arbitrarily small number . then there are no zeros of @xmath206 in the region @xmath251 as long as @xmath96 is sufficiently large . [ cor : fixedzerofree ] this follows immediately from the estimate . indeed , since @xmath217 and @xmath252 are bounded for @xmath253 , zeros of @xmath206 in the region @xmath254 necessarily arise from a balance between @xmath255 and a term of uniform size @xmath256 . however , @xmath255 is large compared with @xmath256 in the region where the inequality @xmath257 holds . a second corollary is an immediate consequence of . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath224 . then there is a constant @xmath258 such that the bound @xmath259 holds for sufficiently large @xmath96 . [ cor : alpha ] this follows directly from with the use of the identity @xmath260 . finally , we have the following result concerning the asymptotic behavior of the normalization constants . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath224 . then there is a constant @xmath258 such that the bound @xmath261 holds for sufficiently large @xmath96 . [ thm : gamma ] we give a direct proof of this theorem based on the identity @xmath262 in [ sec : fixedinside ] . however , another proof with a less sharp error estimate may be based upon theorem [ thm : picircle ] because on @xmath47 uniform convergence implies convergence in @xmath241 . thus , since @xmath263 and @xmath264 , @xmath265 , one finds that @xmath266 next , using , we have @xmath267 substitution into completes the alternate proof . at this point it is important to comment that the @xmath0 method we develop below in [ sec : dbarmethodfixed ] yields new formulae for the polynomial @xmath93 ( see , for example , ) . the formulae are semi - explicit , in that they are written in terms of the solution of a @xmath0 problem ( or , equivalently , in terms of the solution of an integral equation ) . this @xmath0 problem is arrived at after a sequence of explicit transformations , and we prove that this problem has a unique solution , which possesses an asymptotic expansion for @xmath268 . in general , the terms in this expansion can be estimated ( from above ) . such estimations give rise to the general results described in this subsection . however , in the situation that some further information about the weight function @xmath269 is known , it is frequently possible to obtain much more precise information about the terms in the asymptotic expansion . to illustrate what can be obtained from an analysis of the terms of the expansion , we consider in the following subsection a slightly more specific family of weights , and present a rather complete description of the pointwise asymptotic behavior of the polynomials . while the estimate allows one to bound the recurrence coefficients , it does not provide an asymptotic description of the polynomial @xmath206 for @xmath3 bounded within the unit circle . in particular , is insufficient for deducing the location of the zeros . with further assumptions on the regularity of @xmath123 we can extract a leading term that paves the way for further analysis of @xmath206 outside the zero - free region , but within the unit disk . suppose that @xmath222 where @xmath223 is of class @xmath126 with @xmath270 . suppose further that @xmath271 is piecewise continuous with @xmath272 jump discontinuities at points @xmath273 , of magnitudes @xmath274 let @xmath271 have one lipschitz continuous derivative between consecutive jump discontinuities . then , for each @xmath35 , @xmath275 , and @xmath250 , the estimate @xmath276 holds with @xmath277 for all @xmath96 sufficiently large . [ thm : piinsidejumps ] note that @xmath278 is a rational function of @xmath3 with poles at the @xmath279 points of discontinuity of @xmath271 on the unit circle , and with @xmath280 zeros which may lie anywhere in the complex plane , and fluctuate about as @xmath96 is varied . with this result , we can completely characterize the zeros of @xmath206 under the same assumptions on @xmath123 . the simplest example of orthogonal polynomials on the unit circle is of course the case @xmath281 , in which case @xmath282 for all @xmath167 . here we see that all zeros of @xmath206 lie exactly at @xmath91 . in particular , the zeros avoid the unit circle @xmath136 . this situation is typical for strictly positive analytic weights , in which case it is known that the zeros of @xmath206 asymptotically lie within a smaller disk @xmath283 . here , the nearest singularity @xmath284 to the unit circle of the analytic continuation through @xmath136 of the function @xmath216 from the domain @xmath135 determines the radius @xmath245 by @xmath285 . however , such confinement of the zeros within the circle is no longer typical once one leaves the analytic class . for example , discontinuities in any derivatives of @xmath74 make it possible for at most a finite number of zeros to be bounded away from the unit circle while all remaining zeros converge to the unit circle , as the following corollaries of theorem [ thm : piinsidejumps ] show . for each @xmath286 let @xmath287 [ cor : jumpzeroscircle ] assume the same hypotheses as in theorem [ thm : piinsidejumps ] . let @xmath288 denote the annulus @xmath289 then , for each @xmath275 there is some @xmath286 such that the region @xmath290 contains no zeros of @xmath206 for sufficiently large @xmath96 . for each @xmath286 , the zeros of @xmath206 in the region @xmath291 satisfy @xmath292 and @xmath293 modulo @xmath294 , where @xmath295 , and in both cases the error term is uniformly small in the specified region . it follows that the angular spacing between neighboring zeros of @xmath206 in the specified region is @xmath296 . for any fixed @xmath286 , can be rewritten uniformly in the region @xmath297 as @xmath298 and consequently there exists some @xmath299 such that the zeros of @xmath206 in the region @xmath297 asymptotically lie between the two circles @xmath300 . if @xmath286 is sufficiently large , then the region @xmath301 is contained in a disjoint union of small discs centered at the poles @xmath302 of the rational function @xmath278 . in this situation , let @xmath303 denote the component of @xmath301 near the pole @xmath302 . then from we see that the zeros of @xmath206 in the region @xmath304 satisfy @xmath305 where the error term is uniform in the specified region , which indicates that zeros are attracted to a curve that `` bulges '' outward from the circle @xmath306 in a region of angular width proportional to @xmath307 centered at the point @xmath302 , to a maximum radius defined by the equation @xmath308 note that this radius is just within the inner boundary of the zero - free annulus described by corollary [ cor : fixedzerofree ] . the annulus @xmath288 converges toward the unit circle as @xmath102 , and therefore if @xmath3 is a zero of @xmath206 and @xmath96 is large enough , theorem [ thm : piinsidejumps ] gives @xmath309 as @xmath102 because for such @xmath3 , @xmath310 and @xmath311 . the @xmath312 error term is uniformly small for all zeros in @xmath288 . now , let @xmath313 and note that @xmath314 due to the assumptions in force on @xmath124 . since @xmath315 we see that is inconsistent for large enough @xmath96 if @xmath316 . therefore , given @xmath275 , @xmath290 contains no zeros as @xmath102 as long as @xmath317 . for any @xmath286 we now consider those zeros @xmath3 of @xmath206 in the region @xmath318 , in which case we may divide through in by @xmath278 to obtain @xmath319 consistency requires that @xmath320 , and then since the @xmath96-independent exponential factors are continuous up to the unit circle and @xmath288 is converging to the unit circle , we may replace these factors by their limiting values on the unit circle without changing the error estimate . therefore , becomes @xmath321 as @xmath102 uniformly for those zeros @xmath3 of @xmath206 that lie in the annulus @xmath288 . this proves both and . note that the presence of the outward `` bulges '' in the zero curve near the points of discontinuity of @xmath271 indicates the sharpness of the zero - free region established for more general weights in corollary [ cor : fixedzerofree ] . while most zeros of @xmath206 move toward the unit circle as @xmath102 under the hypotheses of theorem [ thm : piinsidejumps ] , there may be at most @xmath280 zeros further inside the unit circle , which correspond to zeros of @xmath278 . we refer to these as `` spurious zeros '' . assume the same hypotheses as in theorem [ thm : piinsidejumps ] . for each @xmath286 , there exists a @xmath275 , such that the zeros of @xmath206 lying in the disk @xmath322 also lie in the set @xmath323 for sufficiently large @xmath96 . moreover , whenever @xmath324 is a sequence of positive numbers such that @xmath325 as @xmath102 , the zeros of @xmath206 in the disk @xmath326 satisfy @xmath327 as @xmath102 . in particular , @xmath206 has exactly one zero for each zero of @xmath278 in this region , making at most @xmath280 spurious zeros . define a constant @xmath328 by @xmath329 therefore , @xmath330 it then follows easily from that all zeros of @xmath206 lying in the disk where the inequality @xmath331 holds will also lie in the set @xmath323 for large enough @xmath96 whenever @xmath332 . the term in proportional to @xmath333 is @xmath312 as @xmath102 uniformly for @xmath3 in the disk delineated by the inequality @xmath326 with @xmath325 as @xmath102 . therefore , @xmath334 for zeros in this disk , and the one - to - one correspondence of zeros of @xmath278 with spurious zeros of @xmath206 in this region follows from the implicit function theorem . note that the zeros of @xmath278 play an apparently contradictory role in the asymptotics . indeed , zeros of @xmath278 that occur near the unit circle repel zeros of @xmath206 , while each zero of @xmath278 that occurs far enough within the unit circle attracts precisely one zero of @xmath206 . a careful reading of the proof of theorem [ thm : piinsidejumps ] ( see [ sec : fixedinside ] ) shows that the @xmath312 estimate that is stated in can be improved to give a rate of decay , and that even better estimates can be obtained if one does not insist on uniformity . these simple improvements can provide , for example , decay rate information for the error terms in the description of the spurious zeros . we have opted not to give these slightly improved estimates in the interest of simplicity of presentation . to illustrate the detailed asymptotic behavior of the zeros of @xmath206 explained above , we have carried out some numerical experiments . let us fix @xmath279 angles of discontinuity @xmath335 by the formula @xmath336 consider the family of weights @xmath74 given by the formula @xmath337 the positive integer @xmath129 and the real numbers @xmath338 are free parameters . this weight is of the form @xmath339 where @xmath271 has jump discontinuities at the points @xmath340 of magnitudes that can be adjusted by choice of the @xmath341 . one advantage of this family from the point of view of numerical computation is that the fourier coefficients of @xmath74 can be evaluated symbolically . in a package such as mathematica capable of arbitrary precision arithmetic , this leads to the possibility of computing the elements of the toeplitz matrices ( whose minors are assembled to yield the coefficients of the polynomial @xmath206 ) with sufficient accuracy for subsequent numerical computation of the zeros when @xmath96 is large . in practice , we computed the coefficients up to an overall factor by scaling the fourier coefficients making up the toeplitz matrix by @xmath342 . this is necessary to avoid numerical overflow or underflow since according to the strong szeg limit theorem the toeplitz determinant of @xmath339 of size @xmath343 scales as @xmath344 . while the fourier coefficients can be computed symbolically , we obtained the coefficients @xmath345 appearing in the rational function @xmath278 defined by with the help of numerical integration . the mathematica code we wrote to carry out these computations is available from the companion website to this paper : . the code takes as input the number of jump discontinuities , @xmath279 , the order @xmath129 of the derivative experiencing the discontinuites , a vector @xmath346 of length @xmath279 containing the parameters @xmath341 , and the degree @xmath96 of the polynomial @xmath206 . the output is a figure showing the unit circle ( black ) with exterior tick marks at the angles @xmath340 , @xmath347 , the zero - attracting circle @xmath306 ( green ) , the inner boundary circle @xmath348 of the zero - free annulus ( red ) , and the zeros of @xmath278 that occur in a neighborhood of the unit disk ( large lavender dots ) . superimposed on the figure are the zeros of @xmath206 ( small black dots ) . sample output from the program is shown in figure [ fig : thumbnail ] . the first effect we would like to illustrate is the rate of convergence of the zeros of @xmath206 to the unit circle with increasing @xmath96 . see figure [ fig:3jumps ] . the annulus associated with the inequalities @xmath349 is asymptotically zero - free , and the curve @xmath306 asymptotically attracts the zeros near the unit circle . the convergence to the zero - attracting circle is clear . more difficult to discern from the images is the outward `` bulging '' of the zeros near the angles @xmath340 of discontinuity toward the inner boundary of the zero - free region . the imaginative reader can see this effect beginning in the figure corresponding to @xmath350 , but larger values of @xmath96 ( and a rescaling of the figures near the unit circle ) will be necessary to resolve the `` bulging '' completely . zeros of @xmath278 play little role for the polynomials whose zeros are illustrated in figure [ fig:3jumps ] . next , we would like to illustrate the effect zeros of @xmath278 can have on @xmath206 ; this is the phenomenon of spurious zeros . note that in the present case of equally - spaced angles @xmath351 , the function @xmath278 is periodic in @xmath96 with period @xmath279 . in figure [ fig : fluctuation ] we present images corresponding to one period of the function @xmath278 in the case of discontinuities of the second and third derivatives of @xmath123 . + here it is clear that the @xmath280 zeros of @xmath278 fluctuate about rapidly with @xmath96 , and can be either inside the unit circle or outside . each zero of @xmath278 inside the unit circle is an asymptote for exactly one zero of @xmath206 , while those outside the unit circle have little effect on the zeros of @xmath206 . the zeros near the unit circle , either inside or outside , have a repulsive effect on the zeros of @xmath206 . from the images in figure [ fig : fluctuation ] it is not obvious that the zeros of @xmath278 inside the unit disk attract corresponding spurious zeros of @xmath206 in the limit @xmath102 . the images shown in figure [ fig : spurious ] show that this convergence indeed occurs . here , we have used the periodicity of @xmath278 to examine the asymptotic behavior of the zeros of @xmath206 along a periodic subsequence of @xmath96-values along which the zeros of @xmath278 remain fixed . + the effect of a zero of @xmath278 upon those of @xmath206 is the most subtle when it occurs near the unit circle . it should be stressed that the parameters @xmath341 of the weight under consideration can be deformed in a continuous manner such that it may always be arranged that @xmath278 has zeros near the unit circle . we fixed @xmath96 and chose a one - parameter deformation of the @xmath352 in order to continuously tune a zero of @xmath278 through the unit circle from outside to inside . a movie of this deformation is available at , and several consecutive frames of this movie are shown below in figure [ fig : deformation ] . + + here it can be clearly seen that as a zero of @xmath278 enters the unit disk , it initially repels the zeros near the attracting circle of @xmath306 by pushing them inwards . the zeros along the attracting curve then move apart to make way for the incoming zero of @xmath278 . exactly one zero of @xmath206 fails to get out of the way , however , and instead enters the orbit of the moving zero of @xmath278 . as the zero of @xmath278 moves inside the attracting circle , it thus draws with it a spurious zero of @xmath206 . here , we begin the task of proving the theorems stated in [ sec : strongasymptoticspi ] by analyzing the behavior of the matrix @xmath152 solving riemann - hilbert problem [ rhp : m ] for a fixed weight @xmath90 , in the limit @xmath102 . we recall the representation of @xmath74 in terms of @xmath223 . in force , in order that riemann - hilbert problem [ rhp : m ] indeed describes the orthogonal polynomials with respect to @xmath74 , is the following . @xmath124 is a real continuous function on the circle that , for some exponent @xmath163 $ ] and for some constant @xmath258 , satisfies a uniform hlder continuity condition @xmath353 . [ ass : positivehoelder ] this guarantees that @xmath74 is a strictly positive function that also satisfies a hlder continuity condition with the same exponent , but with a possibly different constant @xmath164 . we proceed in several steps . first , let @xmath354 be a new unknown related to @xmath152 as follows : @xmath355 it follows from riemann - hilbert problem [ rhp : m ] that the new unknown @xmath354 tends to the identity matrix as @xmath6 , and that @xmath354 is analytic for @xmath153 , with boundary values on the unit circle related by @xmath356 where @xmath357 ( respectively @xmath358 ) indicates the boundary value taken at the point @xmath3 on the circle from the inside ( respectively outside ) . next , observe the following factorization of the jump condition : @xmath359 to take advantage of this factorization , we introduce two new contours @xmath360 which together with @xmath2 bound two concentric annular domains @xmath361 as shown in figure [ fig : domains ] . we will now need some extension of the function @xmath95 defined for @xmath3 on the unit circle with @xmath362 to the annular domain @xmath363 . to make use of the family of extensions defined in , we now make the following assumption about the weight @xmath74 . the function @xmath124 is of class @xmath364 for some @xmath365 . [ ass : ckm1 ] note that when @xmath366 this assumption is contained in assumption [ ass : positivehoelder ] , but when @xmath367 it provides new information . recall the `` bump '' function @xmath146 with the properties listed in [ sec : notation ] . then , for any integer @xmath368 in the range @xmath369 , and for any @xmath35 , we may apply the extension operator @xmath370 to the function @xmath124 and therefore define a matrix @xmath371 as follows . @xmath372 thus , the factor @xmath373 appearing above is our selected extension of the function @xmath95 from the unit circle to the regions @xmath41 and @xmath39 . unlike @xmath152 and hence @xmath354 , the matrix @xmath374 is not piecewise analytic because the factors relating @xmath371 to @xmath354 in the domains @xmath361 are not analytic . indeed , in view of , the exponent @xmath375 is not an analytic function . note however that it follows from assumption [ ass : ckm1 ] and the analyticity of @xmath152 and hence of @xmath354 for @xmath153 that the matrix @xmath374 is continuous for @xmath376 as long as @xmath377 . in particular , the `` bump '' function factor @xmath378 ensures that @xmath371 is continuous across the circles @xmath360 . at the circle of discontinuity , @xmath2 , the boundary values taken by @xmath374 satisfy the jump condition @xmath379 note that the approach to this problem taken in @xcite , where analyticity of @xmath90 is assumed , amounts to replacing @xmath375 with the analytic extension @xmath380 and omitting the `` bump '' function factor @xmath378 , with the latter being at the cost of an exponentially near - identity jump discontinuity across the inner and outer circles @xmath360 . next , we may remove the jump discontinuity along the unit circle by introducing a model matrix @xmath381 that is analytic for @xmath153 , tends to the identity matrix @xmath382 as @xmath6 , and that takes on the unit circle continuous boundary values @xmath383 ( respectively @xmath384 ) from the inside ( respectively outside ) that are related by @xmath385 such a matrix can be found in closed form . setting @xmath386 for @xmath218 and @xmath387 for @xmath135 , one may equivalently seek a matrix @xmath388 that is analytic for @xmath153 , tends to the identity matrix as @xmath6 , and that takes on the unit circle continuous boundary values @xmath389 ( respectively @xmath390 ) from the inside ( respectively outside ) that are related by the diagonal jump condition @xmath391 where @xmath392 denotes the pauli matrix @xmath393 clearly , we may seek @xmath388 as a diagonal matrix . assumption [ ass : positivehoelder ] guarantees that @xmath394 is a well - defined on the circle that satisfies a uniform hlder continuity condition with exponent @xmath163 $ ] , and therefore we may obtain a matrix @xmath388 with the aforementioned properties in the explicit form @xmath395 where @xmath216 is the szeg function associated with the weight @xmath74 as defined in . going back to @xmath381 , the model matrix we will use to remove the jump discontinuity in @xmath374 for @xmath73 is defined by the explicit formula : @xmath396 to actually remove the discontinuity , we introduce a new matrix function @xmath397 defined for @xmath376 by the formula @xmath398 by assumption [ ass : positivehoelder ] and assumption [ ass : ckm1 ] , the matrix @xmath397 is continuous throughout the two regions @xmath399 and @xmath400 . moreover , a continuous extension to @xmath73 is possible because @xmath371 and @xmath381 satisfy the same jump condition at @xmath401 . thus , we see that for @xmath377 , and for any @xmath35 , the matrix function @xmath397 defined by may be viewed as a continuous function on the whole plane with polar coordinates @xmath125 and @xmath402 . at this point , we can summarize the explicit transformations we have introduced and relate @xmath397 directly back to @xmath152 . combining , , , and , we have by definition @xmath403 where @xmath58 . for @xmath404 and @xmath405 we have @xmath406 , and in these regions the matrix @xmath397 clearly inherits analyticity from @xmath152 ; in other words , in these regions @xmath397 is a smooth function of the combination @xmath58 . however , for @xmath407 , the matrix @xmath408 is certainly not analytic . in order to measure the deviation from analyticity in the case when @xmath409 , we introduce a further assumption on @xmath74 . the function @xmath124 is of class @xmath126 , that is , the function @xmath410 is lipschitz continuous . [ ass : ps ] note that since @xmath221 , this condition implies in particular that @xmath123 is lipschitz , and thus the hlder continuity part of assumption [ ass : positivehoelder ] is subsumed . with assumption [ ass : ps ] in force , we may compute the @xmath0-derivative of @xmath397 for @xmath376 and for all integer @xmath368 in the range @xmath369 . differentiation of yields @xmath411 \end{array}\right)\,,&\hspace{0.2 in}0\le r<1\\\\ \displaystyle{\bf m}^n(z)\left(\begin{array}{cc } 0 & 0 \\\\ s_\phi(z)^{-1}{{\overline{\partial}}}\left[b(\log(r)/\epsilon)e^{e_mv(r,\theta)}\right ] & 0\end{array } \right)\ , , & \hspace{0.2 in}r>1\ , , \end{array}\right . \label{eq : dbarem}\ ] ] for almost all @xmath71 , and then elimination of @xmath152 in terms of @xmath397 using again gives @xmath412 where the matrix @xmath413 is given for @xmath376 and almost all @xmath71 by the explicit formula @xmath414\\\\ 0 & 0\end{array}\right)\ , , & \hspace{0.2 in } 0\le r < 1\\\\ \displaystyle \left(\begin{array}{cc } 0 & 0 \\\\ z^{-n}s_\phi(z)^{-2}{{\overline{\partial}}}\left[b(\log(r)/\epsilon)e^{e_mv(r,\theta)}\right ] & 0 \end{array}\right)\,,&\hspace{0.2 in}r>1\ , . \label{eq : wdef}\ ] ] in particular , we see that the matrix @xmath397 is a solution of the following @xmath0-problem . find a @xmath107 matrix @xmath415 with the properties : * * smoothness . * @xmath415 is a lipschitz continuous function throughout @xmath54 . * * deviation from analyticity . * the relation @xmath416 holds for all points in @xmath54 with the exception of a set of measure zero . the matrix @xmath413 is defined almost everywhere by and is essentially compactly supported . * * normalization . * the matrix @xmath415 is normalized at @xmath417 as follows : @xmath418 [ dbp : e ] in writing down this @xmath0-problem , we have focused on just a few specific properties of the matrix @xmath397 . however , it is important that in doing so , we have not introduced any spurious solutions . suppose that @xmath339 where @xmath223 is of class @xmath126 for some @xmath365 . then for all @xmath88 , for @xmath419 , and for all @xmath35 , the matrix @xmath420 is well - defined almost everywhere by and @xmath0 problem [ dbp : e ] has a unique solution , namely @xmath421 . [ prop : dbaresolve ] the existence of a solution follows from and the existence of @xmath152 for @xmath422 . to establish the uniqueness we first consider the determinant of any solution of @xmath0 problem [ dbp : e ] . clearly , @xmath423 is a lipschitz continuous function that tends to @xmath190 as @xmath6 . moreover , the relation @xmath424 holds almost everywhere , and thus by we see that @xmath425 holds almost everywhere in the plane . it follows that @xmath426 is not only lipschitz continuous , but is in fact an entire function of @xmath58 that tends to @xmath190 as @xmath6 . therefore from liouville s theorem we see that @xmath427 . next , consider the matrix ratio of any two solutions @xmath415 and @xmath428 of @xmath0 problem [ dbp : e ] ; this is the matrix @xmath429 defined by @xmath430 since @xmath431 , it follows that @xmath432 is lipschitz continuous throughout the plane . by direct calculation , we have @xmath433 holding almost everywhere in the plane . it follows that @xmath429 is an entire function of @xmath58 that tends to the identity matrix as @xmath6 , so again by liouville s theorem we get @xmath434 , or equivalently @xmath435 . the unique solution of @xmath0 problem [ dbp : e ] can also be expressed as a solution of an integral equation with cauchy kernel . suppose that @xmath339 where @xmath223 is of class @xmath126 for some @xmath365 . then for all @xmath88 , for @xmath419 , and for all @xmath35 , the matrix @xmath413 is well - defined almost everywhere by and the corresponding solution @xmath436 of @xmath0 problem [ dbp : e ] satisfies the integral equation @xmath437 where @xmath58 , @xmath438 , and @xmath439 is a positive area element @xmath440 . the integral is taken over the entire plane . [ prop : dbareinteqn ] recall that the cauchy kernel is a fundamental solution for the @xmath0 operator . in the relation we may replace @xmath441 by @xmath442 $ ] ; multiplying by the cauchy kernel and integrating over the whole plane gives the identity @xmath443}{z'-z}\,da ' = -\frac{1}{\pi}\int\int \frac{{\bf h}^n_{m,\epsilon}(r',\theta'){\bf w}^n_{m,\epsilon}(r',\theta')}{z'-z}\,da'\,.\ ] ] on the left - hand side the @xmath0 operator differentiates with respect to the primed variables . since the cauchy kernel is absolutely integrable , we may evaluate the integral on the left - hand side by replacing the domain of integration by the region @xmath444 and subsequently taking the limit @xmath445 . for each positive @xmath446 we may apply stokes theorem and use the facts that @xmath447=0 $ ] for @xmath448 and that @xmath449 tends to zero as @xmath450 to evaluate the integral over the region @xmath451 in terms of a line integral over the boundary . thus we have @xmath452 where @xmath453 is an arc length element . from the continuity of @xmath408 the integral equation with @xmath421 follows . being as knowledge of the matrix @xmath408 is equivalent to knowledge of @xmath152 and hence of the polynomial of degree @xmath96 in the system of polynomials orthogonal on the circle with respect to @xmath90 , we would like to use the integral equation to characterize @xmath397 . there is a difficulty in that , while existence of solutions for is not an issue , one does not automatically have uniqueness . however , it turns out that if the parameter @xmath96 is sufficiently large , then the integral equation defines a contraction mapping and thus may be solved by iteration yielding a unique solution in the form of a neumann series . in this connection , we can also obtain from asymptotic information about the matrix @xmath397 , and consequently of the orthogonal polynomial @xmath206 , in the limit @xmath102 . in order to study , it is useful to characterize the family of matrix functions @xmath413 more concretely . suppose that @xmath223 is a real function of class @xmath126 for some @xmath221 , that @xmath368 is an integer satisfying @xmath377 , and that @xmath35 is fixed . let the integer @xmath454 be defined as @xmath455 . then , the matrix function @xmath456 is of class @xmath457 if @xmath458 , and of class @xmath459 if @xmath460 . moreover , if @xmath461 and @xmath462 are nonnegative integers such that @xmath463 , then there is a constant @xmath464 such that for all @xmath96 the estimate @xmath465 holds throughout the region @xmath466 containing the essential support of @xmath413 . [ prop : wbound ] @xmath420 vanishes identically outside of the annulus @xmath466 . in the disjoint regions @xmath399 and @xmath400 , the matrix function @xmath420 is infinitely differentiable with respect to @xmath467 , and the issue is the continuity of these derivatives at @xmath73 . the relation implies that for each fixed @xmath71 , @xmath468 is proportional to @xmath469 near @xmath73 ( where @xmath470 holds ) , and thus all derivatives of @xmath413 with respect to @xmath467 through order @xmath471 are lipschitz continuous at @xmath73 , and the derivative @xmath472 remains bounded as @xmath70 , but experiences a jump discontinuity at @xmath73 . on the other hand , if @xmath376 is fixed , then from and , the matrix @xmath473 depends analytically on derivatives @xmath474 for @xmath475 . since @xmath124 is of class @xmath126 , all derivatives of @xmath476 with respect to @xmath71 through order @xmath477 will be lipschitz continuous , while the derivative @xmath478 will be defined for almost all @xmath71 and will be uniformly bounded . to have all mixed partial derivatives of total order at most @xmath479 to be lipschitz continuous , it is sufficient to have both @xmath480 and @xmath481 . if for @xmath377 these inequalities force @xmath460 , then no derivatives of @xmath420 may be taken at all , but @xmath420 is uniformly bounded and compactly supported in the annulus @xmath466 , that is , @xmath482 . if @xmath483 , then we learn that @xmath484 . now there are absolute constants @xmath485 such that @xmath486 where @xmath487\,.\ ] ] generally , the derivatives indexed by @xmath488 and @xmath489 are uniformly bounded throughout the regions @xmath490 and @xmath491 by a constant multiple of @xmath492 . furthermore , using , the derivative @xmath493 is uniformly bounded by a constant multiple of @xmath494 throughout the region @xmath466 . therefore setting @xmath495 , an inequality of the form @xmath496 holds in the region @xmath466 , where @xmath497 is a constant . finally , since @xmath498 , the inequality follows , where @xmath499 . an important part of our analysis will be the estimation of certain two - dimensional laplace - type integrals with cauchy kernels . the main workhorse in this connection is the following lemma . let @xmath35 and @xmath500 be fixed constants . then there exists a corresponding constant @xmath501 such that the estimate ( note that @xmath438 ) @xmath502 holds for sufficiently large @xmath96 . moreover , for each @xmath503 , there exists a constant @xmath504 such that the estimate @xmath505 holds for sufficiently large @xmath96 . [ lem : keybound ] as @xmath506 varies over @xmath47 , the minimum value of @xmath507 is achieved at @xmath508 , and we thus have @xmath509 , which implies the inequality @xmath510 let @xmath511 be a positive constant . clearly , there is another positive constant @xmath512 depending on @xmath511 but not on @xmath467 or @xmath513 such that @xmath514 because the condition @xmath515 also bounds @xmath516 away from zero . furthermore , if @xmath511 is sufficiently small , there is a positive constant @xmath517 depending on @xmath511 but not on @xmath467 or @xmath513 such that @xmath518 to establish , note that for @xmath511 sufficiently small the condition @xmath519 implies that @xmath520 . assuming without loss of generality that @xmath521 , we use the following estimates of the integrand . for @xmath522 we use the estimate @xmath523 which follows from the triangle inequality applied to the identity @xmath524 . for @xmath525 we use the estimate @xmath526 which follows from the diagram shown in figure [ fig : trigonometry ] . finally , for @xmath527 we use the estimate @xmath528 which follows from the law of cosines because @xmath529 : @xmath530 ( again , see figure [ fig : trigonometry ] ) . combining these estimates , we have @xmath531 this is clearly bounded above by a constant multiple of @xmath532 for @xmath511 sufficiently small , and from this follows as well ( adding a constant inside the logarithm keeps the bound positive away from the singularity as long as @xmath174 , and for later purposes it is convenient to take the additive constant to be @xmath96 ) . now we estimate the integral over @xmath513 . using @xmath533 , along with and , and changing the integration variable from @xmath513 to @xmath534 , we get @xmath535 where @xmath536 finally , we have @xmath537 which is finite and independent of @xmath96 and @xmath467 because @xmath500 , while @xmath538 and the last integral is bounded independently of @xmath96 and @xmath467 , since by cauchy - schwarz , @xmath539^{1/2 } \left[\int_{-\infty}^\infty ds \left(\log\left(1+\frac{1}{\|s\|}\right)\right)^2\right]^{1/2}\ ] ] with both factors being finite . thus , an upper bound for the integral of interest is proportional to @xmath540 in general , which proves . if @xmath503 and @xmath541 , then it is not necessary to divide the integration into sets @xmath542 and @xmath543 , and the bound can be used over the whole range of integration in which case the upper bound is then proportional to @xmath544 which proves . with these results in hand , we can formulate and prove the following . [ prop : hbound ] suppose that @xmath339 where @xmath223 is of class @xmath126 for some @xmath365 . let the integer @xmath368 lie in the range @xmath377 and fix @xmath35 . define the integer @xmath545 . then , for all @xmath167 the matrix @xmath413 is well - defined almost everywhere by , and for all @xmath96 sufficiently large , @xmath408 is given by a neumann series @xmath546 which converges in the norm @xmath547 , where the double integral operator @xmath548 is defined by @xmath549 in particular , if @xmath460 then @xmath550 lies in the space @xmath114 , and if @xmath458 then @xmath551 lies in the space @xmath552 and @xmath553 is finite . for all integer @xmath118 in the range @xmath554 , the following estimates hold for sufficiently large @xmath96 : @xmath555 @xmath556 where @xmath557 is a constant . furthermore , for each @xmath503 and for all integer @xmath118 in the range @xmath554 , the following estimates hold for sufficiently large @xmath96 : @xmath558\right\\| \le \tilde{c}^{(p)}_{m,\rho}\frac{1}{n^{m - p}}\ , , \label{eq : hmiaway}\ ] ] @xmath559\right\\| \le \tilde{c}^{(p)2}_{m,\rho}\frac{\log(n)}{n^{2m-2p}}\ , , \label{eq : hmimwiaway}\ ] ] where @xmath560 is a constant . fix @xmath221 and @xmath368 in the range @xmath377 , and set @xmath561 . let @xmath118 be a nonnegative integer satisfying @xmath562 . if @xmath115 , suppose that @xmath563 is matrix function of class @xmath116 with all derivatives of total order no greater than @xmath118 uniformly bounded in the whole plane , and if @xmath112 , suppose that @xmath563 is of class @xmath114 . for such @xmath563 we recall the norm , where the cartesian coordinates @xmath34 and @xmath111 are connected to the polar coordinates @xmath467 and @xmath71 in the usual way : @xmath564 and @xmath565 . because for @xmath566 , @xmath567 for @xmath568 and @xmath569 with @xmath570 , we have @xmath571\,da ' \\\\&=&\displaystyle -\frac{1}{\pi}\int\int\frac{\partial^{\alpha+\beta } } { \partial x^{\prime\alpha}\partial y^{\prime\beta } } \left[{\bf f}(r',\theta'){\bf w}_{m,\epsilon}^n(r',\theta')\right ] \frac{da'}{z'-z}\ , . \end{array } \label{eq : ibp1}\ ] ] note that in order to integrate by parts in for all @xmath461 and @xmath462 of interest , we must have @xmath572 . now , since @xmath420 is compactly supported in the annulus @xmath466 , @xmath573\right\\| & \le & \displaystyle \mathop{\sum_{\alpha'+\alpha''=\alpha}}_{\beta'+\beta''=\beta } \left[\sup_{\mathbb{r}^2}\left\\|\frac{\partial^{\alpha'+\beta ' } } { \partial x^{\alpha'}\partial y^{\beta'}}{\bf f}(r,\theta)\right\\|\right ] \cdot \left\\|\frac{\partial^{\alpha''+\beta''}}{\partial x^{\alpha '' } \partial y^{\alpha''}}{\bf w}_{m,\epsilon}^n(r,\theta)\right\\|\\\\ & \le & \displaystyle \\|\|{\bf f}\|\\|_p\cdot \sum_{\alpha''=0}^\alpha\sum_{\beta''=0}^\beta \left\\|\frac{\partial^{\alpha''+\beta''}}{\partial x^{\alpha '' } \partial y^{\beta''}}{\bf w}_{m,\epsilon}^n(r,\theta)\right\\|\\\\ & \le & \displaystyle \\|\|{\bf f}\|\\|_p\sum_{\alpha'+\beta'\le \alpha+\beta } k_{m,\epsilon}^{(\alpha',\beta')}\left\\|\frac{\partial^{\alpha'+\beta ' } } { \partial r^{\alpha'}\partial \theta^{\beta'}}{\bf w}^n_{m,\epsilon}(r,\theta)\right\\|\ , , \end{array}\ ] ] where @xmath574 are some positive constants . using proposition [ prop : wbound ] , we then find that @xmath575\right\\| \le k_{m,\epsilon}^{(\alpha+\beta)}\cdot \\|\|{\bf f}\|\\|_pe^{-n\|\log(r)\|}\|\log(r)\|^{m-1-(\alpha+\beta)}\sum_{j=0}^{\alpha+\beta}n^j\|\log(r)\|^j\ ] ] where @xmath576 is some constant . finally , we arrive at the estimate @xmath577 where @xmath578 is another constant . set @xmath579 , and note that @xmath500 since as @xmath580 ranges from @xmath581 to @xmath118 , @xmath582 ranges from @xmath583 to @xmath368 , and we have @xmath584 . lemma [ lem : keybound ] may thus be applied to each integral on the right - hand side of , with the result that @xmath585 for some constant @xmath557 and @xmath96 sufficiently large . we note in passing that in order for to provide control of the operator @xmath548 , we need to have @xmath586 . the two restrictions in force on @xmath118 , namely @xmath586 and @xmath587 , have been expressed in the statement of the proposition as the inequality @xmath562 . if one restricts attention to those @xmath58 for which @xmath588 , then lemma [ lem : keybound ] implies the inequality @xmath589 where @xmath590 is a constant and @xmath96 is sufficiently large . from it is clear that if @xmath96 is sufficiently large , the double integral operator @xmath591 defined by the formula and acting in the integral equation thus defines a contraction mapping in the space @xmath116 equipped with the norm @xmath592 , or in the space @xmath114 if @xmath112 . this implies that there is a unique solution of in this space that may be found by iteration resulting in the @xmath592-convergent neumann series @xmath593 in particular , choosing @xmath112 one sees that the neumann series furnishes a unique solution of the integral equation in the space @xmath114 . since proposition [ prop : dbareinteqn ] guarantees that the matrix @xmath397 is a known solution of the integral equation that is ( in particular ) uniformly bounded in the plane , we may identify it with the neumann series for @xmath96 sufficiently large . from this point forward in our proof , we assume that @xmath96 is indeed large enough for this to be the case . since the same neumann series also converges in the norm @xmath592 where @xmath118 may be taken to be as large as @xmath454 , we also learn that if @xmath458 then @xmath594 lies in the space @xmath552 and that @xmath553 is finite . from , taking @xmath592 norms and using , we see that @xmath595 if @xmath96 is large enough that @xmath596 exceeds @xmath597 . since according to the integral equation satisfied by @xmath408 we have @xmath598 we may take norms and use and to learn that @xmath599 holds for sufficiently large @xmath96 . the proof of the estimates and is complete upon appropriate redefinition of the constant @xmath600 . note that implies that the upper bounds in and can be reduced by a factor of @xmath601 if in each case the supremum on the left - hand side is taken over only those values of @xmath467 satisfying @xmath602 for a fixed @xmath503 . this completes the proof of the estimates and upon appropriate redefinition of the constant @xmath603 . to uniformly control @xmath118 derivatives of @xmath604 , proposition [ prop : hbound ] requires that @xmath368 should lie in the range @xmath605 , and therefore in order for there to exist suitable values of @xmath368 , @xmath223 should be of class @xmath126 for some @xmath236 . also , note that the utility of the estimates and is that the matrix @xmath606 is off - diagonal , so the diagonal matrix elements of @xmath607 experience more rapid decay than do the off - diagonal elements . if we solve for @xmath152 in terms of @xmath608 , then since @xmath609 , the monic polynomial of degree @xmath96 in the system of polynomials orthogonal with respect to the inner product @xmath610 defined by , we can easily obtain from proposition [ prop : hbound ] asymptotic formulae for @xmath206 and its derivatives , valid for large @xmath96 , with uniform error estimates . in the region @xmath400 we have for each @xmath35 and for each @xmath611 the exact representation @xmath612 here we have used to write @xmath613 for @xmath614 . equivalently , @xmath615 - z^{-n}h_{m,\epsilon,12}^n(r,\theta)b(\log(r)/\epsilon)e^{e_mv(r,\theta ) -2n(z)}\,,\hspace{0.2 in}r\ge 1\ , . \label{eq : pioutsideequation}\ ] ] if @xmath225 is fixed , then we may choose @xmath35 small enough that @xmath406 whenever @xmath616 . in this case , we have simply latexmath:[\[\pi_n(z)z^{-n}e^{-n(z)}-1 = h_{m,\epsilon,11}^n(r,\theta)-1\,,\hspace{0.2 in } @xmath409 . in this case , using proposition [ prop : hbound ] ( specifically recalling the estimate and the fact that @xmath618 is an off - diagonal matrix ) , we see that for some constant @xmath619 , @xmath620 now as the combination @xmath621 is a function of @xmath3 that is analytic in the region @xmath135 and decaying to zero as @xmath6 , we may express arbitrary derivatives of it as cauchy integrals : @xmath622 = -\frac{p!}{2\pi i}\oint_{\|s\|=\rho}\frac{\pi_n(s)s^{-n}e^{-n(s)}-1}{(s - z)^p } \,ds\,,\ ] ] where the contour of integration is oriented in the counterclockwise direction , and @xmath623 . using to bound the integrand then gives a uniform bound of the same order of magnitude for derivatives over regions bounded away from the circle @xmath624 , which can be taken arbitrarily close to the unit circle . this proves , and completes the proof of theorem [ thm : pioutside ] . theorem [ thm : picircle ] follows from the estimate by noting that the error is an analytic function of @xmath3 in the exterior domain @xmath135 that decays as @xmath6 , and therefore implies the more general result stated in theorem [ thm : picircle ] via the maximum modulus principle . to prove , we fix any positive value of @xmath626 and consider @xmath627 in which case @xmath470 and therefore implies the following formula : @xmath622 = \partial^p\left[h_{m,\epsilon,11}^n(r,\theta)-1\right ] -\partial^p\left[z^{-n}h_{m,\epsilon,12}^n(r,\theta ) e^{e_mv(r,\theta)-2n(z)}\right]\,,\hspace{0.2 in } 1\le r\le 2^{\epsilon/2}\ , . \label{eq : intermediate1}\ ] ] we remind the reader that estimates on derivatives like are valid for @xmath605 ( see the remark at the end of [ sec : dbarmethodfixed ] ) . using the estimate from proposition [ prop : hbound ] and noting that @xmath413 is an off - diagonal matrix , we see that for some constant @xmath232 , @xmath628\right\|\le k_p\frac{\log(n)^2}{n^{2m-2p}}\,,\ ] ] if @xmath629 . on the other hand , the dominant contributions actually come from those terms in the second member of the right - hand side of in which none of the @xmath118 derivatives fall on the exponential factor @xmath630 ( which has @xmath631 uniformly bounded derivatives ) . since @xmath614 , it suffices to estimate @xmath632 with the use of the inequality in proposition [ prop : hbound ] . taking @xmath633 for the best possible decay estimate then gives @xmath634\right\| \le k_p\frac{\log(n)}{n^{k-2p}}\,,\ ] ] where @xmath232 is a constant . this proves , upon taking the limit @xmath635 and writing the @xmath3 derivatives in terms of @xmath71 ( differentiation commutes with the limit process ) . using and the fact ( see ) that @xmath636 , we have the exact representation : @xmath637 and , whenever @xmath218 , @xmath638 where has been used , and @xmath58 . to prove theorem [ thm : gamma ] , we simply apply from proposition [ prop : hbound ] in the case @xmath112 and @xmath409 to the identity . this immediately yields and completes the proof . the proof of theorem [ thm : piinside ] is based on a similar analysis of . recalling that @xmath639 , one can choose @xmath640 and then @xmath470 in for @xmath254 . the estimate then follows by taking @xmath409 , and using from proposition [ prop : hbound ] in the case @xmath112 . similarly , choosing @xmath641 , we have @xmath406 in for @xmath642 . again taking @xmath409 , one obtains by using from proposition [ prop : hbound ] with @xmath112 . this completes the proof of theorem [ thm : piinside ] . the rest of this section will be devoted to the proof of theorem [ thm : piinsidejumps ] . we begin with for @xmath409 , a formula that is valid for all @xmath3 with @xmath218 . using from proposition [ prop : hbound ] , and keeping the term corresponding to @xmath643 , we arrive at the formula @xmath644 as @xmath102 , where the error term is uniform for @xmath218 . here , @xmath440 is an area element , and in the integral @xmath506 varies over @xmath47 ( the support of @xmath645 is the annulus given by the inequalities @xmath646 ) . in the annulus of support of @xmath645 , we have from that @xmath647\ , . \label{eq : wdbarformula}\ ] ] the fact that @xmath645 is in the range of @xmath0 means that the double integral in can be reduced without approximation to the sum of an explicit contribution and a contour integral . indeed , by the inversion of the @xmath0 operator , @xmath648 where @xmath649 denotes the characteristic function of an interval @xmath650 , and where @xmath651 is function analytic except on the circles @xmath652 and @xmath136 bounding the support of @xmath645 that is chosen to make the right - hand side continuous and decaying as @xmath6 . these latter properties uniquely identify @xmath651 with the cauchy integral @xmath653 where the contour of integration is oriented counterclockwise . note that it is the presence of the bump function @xmath378 that makes @xmath651 continuous at @xmath652 . unfortunately , this interesting formula , while apparently simpler than a double integral , is not as useful for asymptotic analysis as the alternative approach we now follow . continuing our analysis , we carry out the differentiation in in the region @xmath218 with the use of : @xmath654\\\\ & = & \displaystyle z^ne^{-2v_0 - 2\overline{n(1/\overline{z } ) } } e^{e_kv(r,\theta)}\left[{{\overline{\partial}}}b(\log(r)/\epsilon ) + \frac{ie^{i\theta}}{2r}\frac{v^{(k)}(\theta)}{(k-1)!}(-i\log(r))^{k-1 } b(\log(r)/\epsilon)\right]\ , , \end{array}\ ] ] and in the special case that @xmath655 , we have @xmath470 and @xmath656 , so @xmath657 the presence of the @xmath255 factor together with the absolute integrability of the cauchy kernel in two dimensions means that @xmath658 holds as @xmath102 uniformly for all @xmath3 with @xmath218 . therefore using the simpler formula in the integrand and integrating over the smaller annulus @xmath659 introduces an error that is uniformly exponentially small for @xmath218 . using the simple identity @xmath660 we may rewrite in the form @xmath661 our subsequent analysis will be specialized to the case where @xmath271 is piecewise continuous , with jump discontinuities at @xmath272 angles @xmath662 , and is ( at first ) only lipschitz between the points of discontinuity . then , between the points of discontinuity , @xmath663 exists almost everywhere and may be identified with a bounded function . under these circumstances , we may `` integrate by parts '' ( _ i.e. _ apply stokes theorem ) with the following formula @xmath664}{z'-z}\,da ' + \frac{1}{\pi}\int\int\frac{g(r',\theta'){{\overline{\partial}}}f(r',\theta')}{z'-z}\,da'\ , , \label{eq : dbaribp}\ ] ] and the first integral on the right - hand side may be exchanged for a sum of explicit terms and a contour integral as described in the above remark . in and in the rest of the proof , whenever the operator @xmath0 appears in the integrand it acts on the primed variables . to prepare to use this technique , we begin with @xmath665\ , , \label{eq : wintegralsum}\ ] ] where @xmath666 is shorthand for the following terms : @xmath667 and @xmath668 refers to the interval in @xmath47 of initial angle @xmath340 and final angle equal to the point of next jump discontinuity as the circle is traversed in the counterclockwise direction . using , we therefore find @xmath669 where @xmath670(-\log(r'))^k}{z'-z}\,da ' \right]\\\\ k_k(r,\theta)&:=&\displaystyle\frac{i^k}{k!}\sum_{j=1}^\ell \left[-\frac{1}{\pi } \mathop{\mathop{\int\int}_{2^{-\epsilon/2}<r'<1}}_{\theta'\in i_j } \frac{{{\overline{\partial}}}\left[(z')^nh(r',\theta')v^{(k)}(\theta')(-\log(r'))^k\right ] } { z'-z}\,da'\right]\ , . \end{array}\ ] ] the double - integral expression @xmath671 may be reduced to contour integrals as follows : @xmath672 where @xmath673 and @xmath674 . since the @xmath675 norm of the numerator in the first cauchy integral is proportional to @xmath676 , we may also write @xmath677 in the form @xmath678 as @xmath268 where the exponentially small error term is uniform for @xmath218 ( right up to the contour of integration ; this is a consequence of the plemelj - privalov theorem @xcite ) . let us now consider @xmath679 . note that @xmath680&=&\displaystyle \left[v^{(k)}(\theta ) { { \overline{\partial}}}e_kv(r,\theta ) + \frac{iz}{2r^2}v^{(k+1)}(\theta)\right ] h(r,\theta)\\\\ & = & \displaystyle \frac{iz}{2r^2}\left[\frac{v^{(k)}(\theta)^2}{(k-1)!}(-i\log(r))^{k-1 } + v^{(k+1)}(\theta)\right]h(r,\theta ) \ , . \end{array}\ ] ] therefore , inserting this into the integrand for @xmath679 and applying lemma [ lem : keybound ] with @xmath681 to the integrals resulting from the first term above , we find @xmath682 \frac{(z')^nh(r',\theta')v^{(k+1)}(\theta ' ) } { z'-z}\,da'\right]\\\\ & & \displaystyle\,\,\ , + \,\,\ , o\left(\frac{\log(n)}{n^{2k}}\right)\ , , \end{array}\ ] ] where the error is uniformly small as @xmath102 for all @xmath218 . with @xmath663 bounded in each @xmath668 , we could in principle apply lemma [ lem : keybound ] to the remaining integrals . however , this would only give a bound of order @xmath683 , and as @xmath677 will turn out to be ( for most @xmath3 ) of size @xmath684 , and we will want to consider @xmath677 to provide the dominant term , we need to impose additional conditions on @xmath685 in each @xmath668 to see that @xmath679 is indeed subdominant . therefore , we first use to write @xmath679 in the form @xmath686 + o\left(\frac{\log(n)}{n^{2k}}\right)\,.\ ] ] the double integral above is of the same form as the original double integral we are trying to compute ( see ) , but with @xmath129 replaced by @xmath687 ( everwhere it appears explicitly ; in the function @xmath666 , @xmath129 remains @xmath129 ) . to continue the analysis of @xmath679 , we therefore use the further assumption that in each each of the intervals @xmath668 , @xmath688 is a lipschitz continuous function so that @xmath689 exists almost everywhere in @xmath668 and can be identified there with a bounded function . repeating the above steps , we find that @xmath690 and with the help of the identity @xmath691=\frac{iz}{2r^2}\left [ \frac{v^{(k)}(\theta)v^{(k+1)}}{(k-1)!}(-i\log(r))^{k-1 } + v^{(k+2)}(\theta)\right]h(r,\theta)\,,\ ] ] we may apply lemma [ lem : keybound ] to the integrals that result from substituting the above into @xmath692 with @xmath693 ( from the first term above ) and with @xmath694 ( from the second term above ) . consequently , @xmath695 where the error is uniformly small as @xmath102 for all @xmath399 . as before , @xmath696 is given by with @xmath697 given by ( note that in substituting @xmath687 for @xmath129 in these formulae , one leaves @xmath666 alone ) . combining these results , we have shown that @xmath698 holds uniformly for @xmath699 under the assumptions in force on @xmath123 . now in addition to @xmath218 we choose arbitrarily a constant @xmath275 and consider those @xmath3 for which @xmath700 ( this is the interesting case , since according to corollary [ cor : fixedzerofree ] we are excluding a zero - free annulus near the unit circle whenever @xmath701 where @xmath446 is the arbitrary positive parameter in the statement of corollary [ cor : fixedzerofree ] ; note that both @xmath446 and @xmath702 may be taken to be arbitrarily small ) . since the function @xmath703 achieves its maximum value when @xmath704 is proportional to @xmath705 , we have @xmath706 since we are assuming here that @xmath707 , @xmath275 may be chosen small enough that both error terms in may be replaced by @xmath708 as @xmath102 uniformly for @xmath709 . it remains to evaluate the explicit integrals in by laplace s method . letting @xmath710 or @xmath711 we consider @xmath712 expecting the dominant contribution to come from the neighborhood of @xmath713 , we write @xmath714x^n(-\log(x))^a\,dx\,.\ ] ] finding a common denominator and extracting from @xmath715 a factor of @xmath716 , we see that @xmath717\ , . \end{array}\ ] ] if @xmath3 does not approach the point @xmath302 , then the fraction in the integrand of the second integral is easily seen to be bounded by a multiple of @xmath718 that is independent of @xmath96 . more generally , if @xmath3 is allowed to approach the point @xmath302 as @xmath102 , then we note that the integrand is analytic in @xmath34 , and the path of integration may be deformed in such a way that throughout the path of integration @xmath719 is bounded away from zero by a quantity that is proportional to @xmath307 because @xmath709 , and therefore the fraction in the integrand of the second integral is bounded by a multiple of @xmath718 that is proportional to @xmath720 . therefore , @xmath721 where the error is uniformly small for @xmath722 . since @xmath723 , we have @xmath724 . it follows that @xmath725 holds uniformly in the region @xmath726 , and therefore by so does @xmath727 this concludes the proof of theorem [ thm : piinsidejumps ] . in this section , we consider weights of the form @xmath728 where @xmath223 is a given real - valued function of period @xmath75 . the weight varies exponentially according to a parameter @xmath96 , and for each @xmath96 we may associate with @xmath90 the corresponding sequence of monic orthogonal polynomials @xmath729 and normalization constants @xmath730 . properly speaking , these quantities depend on the parameter @xmath96 , and we should invent notation to express this dependence , such as @xmath731 . we will not introduce this cumbersome notation . however , the reader should take note of this dependence . the limit of interest here is to study the behavior of the particular monic polynomial @xmath206 of degree @xmath96 in this system along with its normalization constant @xmath84 , in the limit @xmath102 . thus the large parameter @xmath96 enters simultaneously into the degree of the polynomial and also into the weight , and we are studying the asymptotic behavior along the diagonal of a doubly indexed sequence . the asymptotic behavior in this limit is governed by the function @xmath124 , along with some associated functions . first , recall the analytic function @xmath134 defined by for @xmath614 , which is associated with the negative frequency component of the fourier series of @xmath123 . now define @xmath732 = \theta+\omega(\theta)\ , , \label{eq : kappadefine}\ ] ] where @xmath733 is the periodic function defined in . both functions @xmath734 and @xmath144 are as smooth as @xmath124 is . our asymptotic results in this case are the following . let @xmath230 be a fixed integer . suppose that @xmath735 where @xmath223 is of class @xmath126 for some @xmath736 . if @xmath737 is strictly positive , then for each @xmath738 there is a constant @xmath226 such that @xmath739\right\|\le k_{p , \rho}\frac{\log(n)}{n^{2(k-1 ) } } \end{aligned}\ ] ] holds for all sufficiently large @xmath96 . [ thm : varypiasympoutbasic ] let @xmath230 be a fixed integer . suppose that @xmath735 where @xmath223 is of class @xmath126 for some @xmath736 . if @xmath737 is strictly positive , then there is a constant @xmath232 such that @xmath740\right\|\le k_{p}\frac{\log(n)}{n^{k-2p-1 } } \label{eq : varyingpiasympoutside}\ ] ] holds for all sufficiently large @xmath96 . [ thm : picirclevaryingweights ] as in the fixed weights case ( see ) , a special case of the estimate is the following estimate ( holding under the same conditions ) @xmath741\right\|\le k_p\frac{\log(n)}{n^{k-2p-1}}. \label{eq : varyingpiasympcircle}\ ] ] asymptotically characterizing @xmath93 on the unit circle . once again , the uniform nature of the convergence on the circle allows us to prove the following mean result ( the proof is the same as for the analogous result in the fixed weight case ) . let @xmath230 be a fixed integer . suppose that @xmath735 where @xmath223 is of class @xmath126 with @xmath736 . if @xmath737 is strictly positive , then @xmath742 the next result concerns the asymptotic behavior of the polynomial @xmath93 for @xmath3 inside a closed annular region whose outer boundary is the unit circle . suppose that @xmath735 where @xmath223 is of class @xmath126 for some @xmath707 . if @xmath737 is strictly positive , then for each @xmath245 satisfying @xmath246 there are constants @xmath247 such that the estimates @xmath743 and @xmath744 hold for all @xmath96 sufficiently large . [ thm : varyingcircleattract ] an immediate corollary is that there exists an annulus inside the unit circle that asymptotically contains no zeros . suppose that @xmath735 where @xmath223 is of class @xmath126 for some @xmath707 , and suppose that @xmath737 is strictly positive . let @xmath250 be an arbitrarily small number . then there are no zeros of @xmath206 in the region @xmath745 as long as @xmath96 is sufficiently large . [ cor : varyzerofree ] in the case of a fixed weight , corollary [ cor : fixedzerofree ] established the existence of a zero - free annulus with outer radius 1 , and whose inner radius depended explicitly on @xmath129 , the degree of smoothness . we subsequently considered a family of weights for which we could compute explicitly the behavior of the zeros , and showed that corollary [ cor : fixedzerofree ] is sharp . indeed , for explicit families of weights with @xmath129 degrees of smoothness , some zeros achieve a distance of @xmath746 from the unit circle . in addition , for these examples , a majority of the zeros approach a circle whose radius is @xmath747 . for the case of varying weights , it is our belief that corollary [ cor : varyzerofree ] is similarly sharp . moreover , it is to be expected that for some canonical family of examples constructed such that @xmath748 has jump discontinuities , it should be similarly possible to obtain a very detailed asymptotic description of the zeros . the zero - free region is determined by observing that for there to be a zero of @xmath93 , @xmath749 must be roughly the same size as @xmath750 . this is clearly not true for @xmath3 on the unit circle , and investigating the size of @xmath751 relative to @xmath750 yields the result . for @xmath218 , @xmath749 is small not only because @xmath255 is small , but also because ( as will be clear in [ sec : varyingdetails ] below ) we constructed @xmath752 so that for @xmath399 but @xmath753 sufficiently small , @xmath754 thus our extension of the function @xmath144 plays a role in determining the zero - free region near the unit circle . the quantity @xmath755 appearing in corollary [ cor : varyzerofree ] is strictly positive because @xmath756 , and @xmath737 is strictly positive . it is interesting to note that this condition also guarantees that there are no gaps in the support of the equilibrium measure ( see appendix [ app : potential ] ) . the occurrence of a gap in the support of the equilibrium measure is heralded by the development of a zero of the function @xmath737 . although corollary [ cor : varyzerofree ] would not apply if @xmath737 vanished at some @xmath757 , an intuitive consideration of the set defined in indicates that near @xmath757 , the zeros are pushed away , further from the unit circle . as a gap develops then , one might expect the zeros to accumulate on a contour approaching the unit circle , but with a gap aligned with the gap in the support of the equilibrium measure . we will not carry out such an analysis here , but clearly the methods outlined here can be adapted in this direction . just as theorem [ thm : piinside ] leads to corollary [ cor : alpha ] , from theorem [ thm : varyingcircleattract ] yields the following result . ( recall from the definition that @xmath758 as @xmath759 , and so @xmath760 as @xmath761 . ) suppose that @xmath735 where @xmath223 is of class @xmath126 with @xmath762 . if @xmath737 is strictly positive , then there is a constant @xmath258 such that the estimate latexmath:[\[\label{eq : varyverblunskyas } @xmath96 . [ cor : varyingalpha ] finally , we have the following result concerning the asymptotic behavior of the normalization constant @xmath764 , defined such that @xmath765 , with @xmath74 given in the varying weights case by . suppose that @xmath124 is a real function of class @xmath126 for some @xmath707 . if @xmath737 is strictly positive , then there is a constant @xmath258 such that @xmath766 holds for all @xmath96 sufficiently large . [ thm : varyinggamma ] at first glance , the asymptotic formulae and may appear to be inconsistent with the identity . however , in , @xmath767 is the norming coefficient of the @xmath768-st degree polynomial orthogonal with respect to the @xmath96-dependent weight @xmath769 . to verify in the varying weights case , one would have to compute the asymptotics for @xmath770 as well , rather than merely replacing @xmath96 by @xmath343 in . one of the main points of this paper is that while the limit appropriate for exponentially varying weights of the form lies for the most part beyond the reach of classical techniques applicable for fixed weights ( meaning primarily the approximation of @xmath95 by positive trigonometric polynomials ) , analysis of this limit by means of the @xmath0 steepest descent method presents almost no further difficulty beyond the analysis carried out for fixed weights in [ sec : strong ] . to illustrate the ease with which many of the techniques carry over from the fixed weights case , we now outline the analogous calculations for the varying weight . as in the case of fixed weights , we begin with the matrix @xmath151 solving riemann - hilbert problem [ rhp : m ] and introduce a sequence of explicit transformations . in order that riemann - hilbert problem [ rhp : m ] indeed characterizes the polynomial of degree @xmath96 orthogonal to all lower degree polynomials with respect to the weight , we make the following assumption . @xmath124 is a real continuous function on the circle that , for some exponent @xmath163 $ ] and for some constant @xmath258 , satisfies a uniform hlder continuity condition @xmath353 . [ ass : kappa1 ] now , define @xmath771 in terms of @xmath152 as follows : @xmath772 since @xmath758 as @xmath6 , it is easy to see that as @xmath6 , @xmath773 . moreover , @xmath771 is analytic for @xmath153 , and its boundary values @xmath774 ( respectively @xmath775 ) taken on the circle @xmath47 from the inside ( respectively outside ) satisfy : @xmath776 note that according to assumption [ ass : kappa1 ] , @xmath777 also satisfies a uniform hlder continuity condition with exponent @xmath582 . as in [ sec : strongconvert ] , we have available an algebraic factorization of this jump condition : @xmath778 to take advantage of this factorization we need to extend the functions @xmath779 from @xmath47 to an annulus containing @xmath47 . to have continuity of these extensions , we make the following assumption . the function @xmath124 is of class @xmath364 for some @xmath365 . [ ass : kappa2 ] with this assumption , which provides new information only if @xmath367 , we can extend @xmath779 as follows . writing @xmath780 , we extend the factor @xmath781 analytically as @xmath782 . on the other hand , @xmath142 is a well - defined function on the circle , and therefore we may apply the extension operator @xmath370 defined in to @xmath144 , resulting in an extension of @xmath783 to the domain @xmath784 as a continuous function @xmath785 for any @xmath368 in the range @xmath786 . let @xmath35 be the radius parameter of the annular domains @xmath361 ( see figure [ fig : domains ] ) . recall the `` bump '' function @xmath146 defined in [ sec : notation ] . we define a new matrix unknown @xmath787 by setting @xmath788 note that the presence of the factor @xmath378 ensures that @xmath789 may be continuously extended to the outer boundary of @xmath39 and the inner boundary of @xmath41 ; that is , @xmath789 is a continuous function for @xmath376 . here our analysis rests upon extending @xmath733 from the circle into domains @xmath361 . by contrast in our treatment of the fixed weights case in [ sec : dbarmethodfixed ] it was more convenient to extend the function @xmath123 , which is related to @xmath733 by a cauchy transform . if @xmath790 , then it is easy to see that @xmath791 for all @xmath792 , so the off - diagonal matrix elements in @xmath793 are bounded in magnitude by @xmath794 . therefore , in this case @xmath795 represents an exponentially near - identity transformation when @xmath96 is large and @xmath704 is not too small . we want to ensure that a similar situation prevails when @xmath796 as well . since @xmath377 , the consideration of @xmath796 requires that @xmath707 . in this case assumption [ ass : kappa2 ] implies that @xmath777 is continuously differentiable , and hence so is @xmath733 . the crucial conditions for to be a near - identity transformation when @xmath96 is large are that @xmath797 is strictly positive and that the parameter @xmath626 is chosen small enough , as the following lemma shows . suppose that @xmath733 is a real function of class @xmath49 for some @xmath796 , such that @xmath797 is strictly positive . then there exist constants @xmath798 and @xmath799 such that whenever @xmath800 , @xmath801 and @xmath802 [ lemma : exponentcontrol ] by hypothesis , we have @xmath803 note that @xmath804 . from , we have @xmath805\,,\ ] ] where @xmath806 if @xmath368 is even and @xmath807 if @xmath368 is odd . since @xmath144 has @xmath69 continuous derivatives we have @xmath808 for all @xmath467 satisfying @xmath466 where @xmath35 . therefore , @xmath798 may be chosen sufficiently small that the inequality @xmath800 implies that @xmath809 holds whenever @xmath466 . then , we have @xmath810\,,\hspace{0.2 in } \text{holding for $ 1\le r\le 2^\epsilon$}\ ] ] and @xmath811\,,\hspace{0.2 in } \text{holding for $ 2^{-\epsilon}\le r\le 1$}\,.\ ] ] the estimates and thus both hold with the choice @xmath812 . thus , we are led to propose the following assumption . if @xmath124 is of class @xmath813 , then @xmath737 defined by is strictly positive . [ ass : kappa3 ] this assumption is not an explicit assumption on @xmath123 , however it is possible to give conditions on @xmath123 that are sufficient to make assumption [ ass : kappa3 ] hold . for example , since for @xmath814 , we have @xmath815 where we recall the fourier coefficients of @xmath124 defined by , the condition @xmath816 is sufficient to guarantee that assumption [ ass : kappa3 ] holds . if in fact @xmath817 , then the convexity condition @xmath818 also guarantees that @xmath737 is strictly positive . our proof that the condition implies that @xmath737 is strictly positive makes use of certain aspects of logarithmic potential theory and is given in appendix [ app : potential ] . in appendix [ app : compare ] , we show that neither are the sufficient conditions ( [ eq : tcondition ] ) and ( [ eq : convexity ] ) equivalent , nor does either condition imply the other . rather , the two conditions are independent and thus complement each other . in the approach taken in @xcite , where a particular case of a varying weight in which the function @xmath123 , and hence @xmath777 , is an analytic function , the analytic extension @xmath819 is used , and the `` bump '' function factor @xmath378 is omitted . the latter omission has the effect of introducing exponentially small jump discontinuities in @xmath789 across the circles @xmath360 . since for @xmath611 we have @xmath820 , the matrix @xmath821 satisfies the jump condition @xmath822 to remove this jump discontinuity , introduce one further change of variables : @xmath823 at this point , we can relate @xmath824 directly to @xmath152 . combining , , and , we have by definition @xmath825 given the assumptions in force , this matrix function is clearly continuous throughout the plane . to determine its deviation from being an analytic function in the regions @xmath399 and @xmath400 , we need to control a derivative , and consequently we make the following assumption . the function @xmath124 is of class @xmath126 for some @xmath221 . that is , @xmath410 is lipschitz continuous . [ ass : kappa4 ] then , for @xmath377 , we see by direct calculation that @xmath826 \end{array}\right)\ , , \\\\ \hspace{4 in } \text{for $ 0\le r < 1$}\\\\ \displaystyle e^{nv_0\sigma_3/2 } { \bf m}^n(z)e^{-nv_0\sigma_3/2}\left(\begin{array}{cc } 0 & 0 \\\\ e^{nn(z)}{{\overline{\partial}}}\left[b(\log(r)/\epsilon ) e^{-ine_m\omega(r,\theta)}\right ] & 0 \end{array}\right)\,,\\\\\hspace{4 in } \text{for $ r>1$}\ , . \end{array } \right . \label{eq : dbarj}\ ] ] eliminating @xmath152 in favor of @xmath824 using again yields @xmath827 where @xmath828\\\\ 0 & 0 \end{array}\right)\,,&\hspace{0.2 in}\text{for $ 0\le r<1$}\\\\ \displaystyle\left(\begin{array}{cc } 0 & 0 \\\\ r^{-n}e^{-in\theta}{{\overline{\partial}}}\left[b(\log(r)/\epsilon)e^{-ine_m\omega(r,\theta)}\right ] & 0 \end{array } \right)\,,&\hspace{0.2 in}\text{for $ r>1$}\ , . \end{array } \right . \label{eq : xdef}\ ] ] then it is easy to see that @xmath824 satisfies the following @xmath0 problem . find a @xmath107 matrix @xmath415 with the properties : * * smoothness . * @xmath415 is a lipschitz continuous function throughout @xmath54 . * * deviation from analyticity . * the relation @xmath829 holds for all points in @xmath54 with the exception of a set of measure zero . the matrix @xmath830 is defined almost everywhere by and is essentially compactly supported . * * normalization . * the matrix @xmath415 is normalized at @xmath417 as follows : @xmath831 [ dbp : u ] in exactly the same way that proposition [ prop : dbaresolve ] was proved , we have the following . suppose that @xmath832 where @xmath223 is of class @xmath126 for some @xmath365 . then for all @xmath88 , for @xmath419 , and for all @xmath35 , the matrix @xmath833 is well - defined almost everywhere by and @xmath0 problem [ dbp : u ] has a unique solution , namely @xmath834 . also , the proof of proposition [ prop : dbareinteqn ] carries over to the context of the varying weight in the following form . suppose that @xmath832 where @xmath223 is of class @xmath126 for some @xmath365 . then for all @xmath88 , for @xmath419 , and for all @xmath35 , the matrix @xmath833 is well - defined almost everywhere by and the corresponding solution @xmath835 of @xmath0 problem [ dbp : u ] satisfies the integral equation @xmath836 where @xmath58 , @xmath438 , and @xmath439 is a positive area element @xmath440 . the integral is taken over the entire plane . [ prop : inteqnu ] as in [ sec : stronginteqn ] , it is possible to characterize @xmath837 by analyzing the integral equation as long as @xmath96 is large enough . however , in the context of the varying weight we will require the monotonicity condition expressed in assumption [ ass : kappa3 ] , and that the radius parameter @xmath626 be taken sufficiently small for each admissable given @xmath124 that the conclusion of lemma [ lemma : exponentcontrol ] holds . moreover , the exponential character of the varying weight suggests that , by comparison with the analysis presented in [ sec : stronginteqn ] , this approach is only fruitful in giving the same degree of control on @xmath838 as was achieved for @xmath839 if @xmath123 has more smoothness than was required in proposition [ prop : hbound ] . in particular , we will require that the following assumption holds . the function @xmath124 is of class @xmath126 for some @xmath707 . [ ass : kappa5 ] in other words , to achieve the same convergence rates as in the fixed weight case , one more derivative of @xmath124 will be required in the exponentially varying weight case . as in [ sec : stronginteqn ] , we need some bounds for @xmath830 and its derivatives . suppose that @xmath223 is a real function of class @xmath126 for some @xmath221 for which @xmath797 is strictly positive , that @xmath368 is an integer satisfying @xmath377 , and that @xmath35 is sufficiently small . let the integer @xmath454 be defined as @xmath455 . then , the matrix function @xmath840 is of class @xmath457 if @xmath458 , and of class @xmath459 if @xmath460 . moreover , if @xmath461 and @xmath462 are nonnegative integers such that @xmath463 , then there are constants @xmath799 and @xmath464 such that for all @xmath96 the estimate @xmath841 holds throughout the region @xmath466 containing the essential support of @xmath830 . [ prop : xbound ] this proposition is proved in almost the same way as proposition [ prop : wbound ] . in this case , it is essential to recall lemma [ lemma : exponentcontrol ] which provides the constant @xmath799 and thus the exponential decay of the term @xmath842 for @xmath35 sufficiently small . also , the initial application of the @xmath0 operator in the definition yields a factor of @xmath96 that does not appear in the proof of proposition [ prop : wbound ] . this result should be compared with proposition [ prop : wbound ] . the only important difference between the estimate and is the presence of an additional factor of @xmath96 . the proof of proposition [ prop : hbound ] , with references to proposition [ prop : wbound ] replaced by references to proposition [ prop : xbound ] , applies to the asymptotic estimation of @xmath837 , with the only important difference being an additional factor of @xmath96 . this results in the following . [ prop : jbound ] suppose that @xmath832 where @xmath223 is of class @xmath126 for some @xmath843 . let the integer @xmath368 lie in the range @xmath844 and fix @xmath35 sufficiently small . define the integer @xmath845 . then , for all @xmath167 the matrix @xmath830 is well - defined almost everywhere by , and for all @xmath96 sufficiently large , @xmath838 is given by a neumann series @xmath846 which converges in the norm @xmath847 , where the double integral operator @xmath848 is defined by @xmath849 in particular , if @xmath850 then @xmath851 lies in the space @xmath114 , and if @xmath852 then @xmath853 lies in the space @xmath854 and @xmath855 is finite . for all integer @xmath118 in the range @xmath856 , the following estimates hold for sufficiently large @xmath96 : @xmath857 @xmath858 where @xmath600 is a positive constant . furthermore , for each @xmath503 and for all integer @xmath118 in the range @xmath859 , the following estimates hold for sufficiently large @xmath96 : @xmath860\right\\| \le \tilde{c}^{(p)}_{m,\rho}\frac{1}{n^{m - p-1}}\ , , \label{eq : varyinghmiaway}\ ] ] @xmath861\right\\| \le \tilde{c}^{(p)2}_{m,\rho}\frac{\log(n)}{n^{2m-2p-2}}\ , , \label{eq : varyinghmimwiaway}\ ] ] where @xmath862 is a positive constant . note that in , due to the extra factor of @xmath96 introduced into the estimates by proposition [ prop : xbound ] , convergence of the neumann series in the @xmath592 norm follows provided that @xmath863 , and so we may only consider @xmath864 . in order to uniformly control @xmath118 derivatives of @xmath865 , proposition [ prop : jbound ] requires that @xmath368 should lie in the range @xmath866 , and therefore in order for there to exist suitable values of @xmath368 , @xmath223 should be of class @xmath126 for some @xmath736 . assumption [ ass : kappa3 ] , that @xmath123 is such that @xmath777 is a strictly increasing function of the angle @xmath71 , ties together two crucial aspects of our analysis . first of all , as can be seen from the matrix factors involved in the change of variables between @xmath771 and @xmath867 , the inequality @xmath868 is precisely what makes this a near - identity change of variables in the regions @xmath869 . on the other hand , it would not suffice to replace @xmath737 by another unrelated positive quantity , because we require of whatever extension we choose some degree of vanishing of the @xmath0 derivative at @xmath136 in order to control the @xmath0 problem ( that is , to sufficiently bound the operator @xmath848 ) . this vanishing is built into the extension operators we have defined in by the key property . moreover , _ any _ smooth extension of @xmath777 that has a vanishing @xmath0 derivative on the unit circle will behave similarly near the unit circle . so we conclude that while the analyticity of @xmath777 is not important , the monotonicity of this function is crucial . to make an analogy with the asymptotic analysis of oscillatory exponential integrals , the @xmath0 steepest descent method resembles kelvin s method of stationary phase more than it does the saddle point method . on the other hand , it is important to note that the @xmath0 steepest descent method remains fundamentally a method of deformation into the complex plane , and is not based on integration by parts operator in the plane in as much as this can be considered a generalization of the standard @xmath0 inversion formula ) to establish the existence of an asymptotic expansion . however , such methods are useful in the detailed analysis of individual terms in the expansion . this technique was used , for example , in the proof of theorem [ thm : piinsidejumps ] . ] , the riemann - lebesgue lemma , or related arguments of harmonic analysis . for an approach based on the latter , see varzugin @xcite . the proofs are generally based on expressing @xmath870 in terms of explicit functions and @xmath837 by means of . then , one applies proposition [ prop : jbound ] to control @xmath871 and its derivatives . from we have the following exact representation for @xmath206 valid for @xmath135 : @xmath872\ , , \hspace{0.2 in}\|z\|>1\,.\ ] ] here @xmath626 should be taken to be sufficiently small . it follows that @xmath873 -b(\log(r)/\epsilon)z^{-n}e^{-ine_m\omega(r,\theta ) } j_{m,\epsilon,12}^n(r,\theta)\,,\hspace{0.2 in}\|z\|>1\,.\ ] ] if @xmath225 is fixed , then perhaps by making @xmath35 smaller yet , it can be arranged that @xmath406 whenever @xmath874 . in this case , we have @xmath875 = \partial^p \left[j_{m,\epsilon,11}^n(r,\theta)-1\right]\,,\hspace{0.2 in } the case @xmath633 , and the fact that @xmath830 is off - diagonal , we see that for some constant @xmath328 , @xmath876\right\|\le c\frac{\log(n)}{n^{2(k-2p-1)}}\ , . \label{eq : varyj11outest}\ ] ] now since @xmath877 is analytic for @xmath623 , and tends to zero like @xmath878 as @xmath879 , cauchy s theorem for an exterior domain , together with for @xmath112 , proves . to prove , we fix @xmath626 sufficiently small and consider @xmath627 in which case @xmath470 so that @xmath880 = \partial^p\left[j_{m,\epsilon,11}^n(r,\theta)-1\right]- \partial^p\left[z^{-n}e^{-ine_m\omega(r,\theta)}j_{m,\epsilon,12}^n(r,\theta ) \right]\ , , \hspace{0.2 in } 1\le r\le 2^{\epsilon/2}\ , . \label{eq : intermediate3}\ ] ] again , since @xmath830 is an off - diagonal matrix , we see from that there is a constant @xmath328 such that @xmath881\right\|\le c\frac{\log(n)^2}{n^{2(k-2p-1)}}\,.\ ] ] as in the fixed weights case , the dominant contribution comes from the remaining terms on the right - hand side of . using from proposition [ prop : jbound ] to see that @xmath882 is of order @xmath883 and taking the best case of @xmath633 , we then find that @xmath884\right\| \le k\frac{\log(n)}{n^{k-2p-1}}\,.\ ] ] this proves . now the maximum modulus principle applied to @xmath885 implies . the proof of theorem [ thm : varyinggamma ] is based on the identity @xmath262 . using , we see that @xmath886 and thus theorem [ thm : varyinggamma ] is proved by applying the estimate from proposition [ prop : jbound ] in the case of @xmath112 and @xmath409 . proving theorem [ thm : varyingcircleattract ] begins with the exact formula @xmath887 which follows from using @xmath870 and the fact that @xmath406 for @xmath404 . using from proposition [ prop : jbound ] with @xmath112 and @xmath409 then completes the proof of . for the proof of , one begins with the following formula for @xmath888 , @xmath889 which again follows from , and the fact that @xmath470 for @xmath890 . using from proposition [ prop : jbound ] with @xmath112 and @xmath409 then completes the proof of , and this completes the proof of theorem [ thm : varyingcircleattract ] . during our preparation of this work , we benefited from useful conversations with percy deift , andrei martinez - finkelshtein , paul nevai , and walter van assche . we also thank barry simon for letting us see a preliminary version of his forthcoming monograph @xcite . the research of k. t .- mclaughlin was supported in part by the national science foundation under grants dms-9970328 and dms-0200749 . the work of p. d. miller was also supported in part by the national science foundation , under grant dms-0103909 , and by a grant from the alfred p. sloan foundation . a more general and systematic strategy for extracting asymptotics of @xmath152 in the exponentially varying weight case when @xmath74 is of the form is the following . first introduce a function @xmath29 ( to be determined ) satisfying @xmath891 as @xmath6 and that @xmath892 is analytic for @xmath153 , taking continuous boundary values on the unit circle @xmath2 . then one converts riemann - hilbert problem [ rhp : m ] for @xmath152 into one with identity asymptotics as @xmath6 by the change of variables @xmath893 resulting in a new unknown matrix @xmath894 . in addition to the normalization condition @xmath895 , @xmath894 satisfies the jump condition @xmath896 where the subscript `` @xmath897 '' indicates the boundary value taken from within the circle , and `` @xmath898 '' indicates the boundary value taken from outside . without any loss of generality , we may represent @xmath29 in the form of a complex logarithmic potential @xmath899 where for each @xmath65 we consider the function @xmath900 to be real for @xmath901 sufficiently large and positive real , with the branch cut from the point @xmath83 in the clockwise direction along the unit circle to the negative real axis , and then along the negative real axis to @xmath902 . we also require that @xmath903 in order to satisfy the required normalization condition @xmath904 as @xmath6 . additional conditions may now be placed on @xmath905 , or equivalently on @xmath906 , in order to make the jump condition for @xmath894 asymptotically tractable . one key condition is that the function @xmath906 should be real - valued . from this condition it follows that @xmath907 has a constant imaginary part for @xmath136 , as the following argument shows . when @xmath908 , the identity @xmath909 yields @xmath910 using and differentiating under the integral with respect to @xmath3 for @xmath153 , we obtain @xmath911\,.\ ] ] assuming reality of @xmath906 we can pass the real part under the integral and thus arrive at @xmath912 where @xmath913 . the numerator of the fraction in the integrand is clearly uniformly bounded by a quantity of order @xmath914 , and the minimum value of the denominator is achieved when @xmath915 yielding a minimum value that has the asymptotic expansion @xmath916 as @xmath45 . consequently the fraction is uniformly bounded independent of @xmath626 . moreover , the fraction is easily seen to tend to zero pointwise in the limit @xmath45 for @xmath917 . a dominated convergence argument therefore shows that the limit on the right - hand side of is zero . see also @xcite , where it is also shown that the constant value is exactly @xmath918 . another consequence of assuming that @xmath906 is real comes from noting that @xmath919 it therefore follows that @xmath920 is purely imaginary for @xmath136 . since @xmath921 one is led to seek @xmath906 so that the circle is split into intervals of two different types : * * bands : * for @xmath71 in a band @xmath650 , @xmath922 is real and positive , and @xmath923 where @xmath279 is a real constant ( the same constant for all bands in fact if there is only one band it turns out that @xmath924 ) . thus , in a band the jump condition for @xmath894 takes the form @xmath925 where @xmath926 is a strictly increasing real function . * * gaps : * for @xmath71 in a gap @xmath927 , @xmath928 and we have the strict inequality @xmath929 thus , in a gap , the jump condition for @xmath894 takes the form @xmath930 where @xmath931 is a real constant . the alternative conditions and are exactly the euler - lagrange conditions for the minimization of the weighted logarithmic energy @xmath932:=\int_{-\pi}^\pi\int_{-\pi}^\pi\log\left(\frac{1}{\left\|e^{i\theta}-e^{i\theta'}\right\|}\right)\psi(\theta')\psi(\theta)\,d\theta'\,d\theta + \int_{-\pi}^\pi v(\theta')\psi(\theta')\,d\theta'\ ] ] over all probability measures @xmath933 supported on the unit circle . the constant @xmath279 is the lagrange multiplier introduced to enforce the constraint . the minimizing measure with density @xmath922 is called the _ equilibrium measure_. the connection of this extremal problem with the jump condition through the euler - lagrange variational conditions suggests that logarithmic potential theory plays an important role in the asymptotic theory of orthogonal polynomials for general exponentially varying weights of the form . the use of equilibrium measures for the asymptotic analysis of the riemann - hilbert problem associated to orthogonal polynomials on @xmath934 with analytic exponentially varying weights was carried out in @xcite . for orthogonal polynomials on the unit circle with analytic exponentially varying weights of a specific form , this was done in @xcite . the transformation @xmath935 leads to jump conditions of the form @xmath936 which is of exactly the same form as , and @xmath937 clearly , we also have the normalization condition @xmath938 as @xmath6 . now , if the function @xmath123 is such that the whole unit circle consists of a single band , or equivalently there are no gaps in the support of the equilibrium measure , then in fact @xmath939 as defined in [ sec : varyingconvert ] and the analysis proceeds as in the main body of this paper . however , in the more general context when it is only true that the support of the equilibrium measure consists of a finite number of disjoint intervals on the unit circle a modification of the @xmath0 steepest descent method described in [ sec : varying ] is required . a more general method may still be based on the algebraic factorization of the jump matrix in each band ; however the annuli @xmath361 must be replaced with a system of lens - shaped regions @xmath940 adjacent to each band @xmath650 . since the variational inequalities become less effective near the band edges , a local analysis must be supplied to control the error . when such a method is developed , the conditions or on the function @xmath123 can be dropped as long as it is known that there are only a finite number of gaps in the support of the equilibrium measure . in @xcite it is shown how to carry out such a program in the context of nonanalytic exponentially varying weights on the real line , where a convexity condition is known to guarantee the existence of a single isolated band @xmath941\subset\mathbb{r}$ ] . let us show how the logarithmic potential theory described briefly above leads to the condition guaranteeing that the support of the equilibrium measure is the entire unit circle . if @xmath124 is of class @xmath942 , then for @xmath71 in any gap in the support of the equilibrium measure the real function @xmath943 is also twice differentiable . in fact , a calculation shows that @xmath944 note that this is not a singular integral since @xmath945 is assumed to lie outside the support of the equilibrium measure . let us assume both the condition , and also the existence of a gap in the support of the equilibrium measure ; we will then derive a contradiction . now , @xmath946 is continuous for @xmath947 . therefore , since @xmath948 at both endpoints of the gap ( according to ( [ eq : bandcondition ] ) ) , and @xmath949 strictly in the interior of the gap ( according to ( [ eq : gapcondition ] ) ) , there must be a point @xmath71 in the gap at which @xmath950 . however , since @xmath951 for all angles @xmath506 , we see that @xmath952 which is negative in view of the assumption , thus establishing the desired contradiction . in this appendix , we illustrate by concrete examples how the two conditions and , while both sufficient for the prevention of gaps in the support of the equilibrium measure , are completely independent . thus neither condition implies the other . first , consider the example @xmath953 where @xmath365 and @xmath954 are parameters . a direct calculation shows that @xmath955 since @xmath956 for all @xmath365 , we see that if @xmath957 then condition ( [ eq : tcondition ] ) is satisfied but ( [ eq : convexity ] ) is not . next , consider for @xmath286 and @xmath35 the example @xmath958 on one hand , we have @xmath959 and so @xmath960 on the other hand , direct calculation of the fourier coefficients gives @xmath961 thus , we may estimate the sum in the condition ( [ eq : tcondition ] ) as follows : @xmath962 since @xmath963 for @xmath964 . if the product @xmath965 is held fixed this lower bound can be made arbitrarily large ( and in particular larger than one ) simply by taking @xmath626 sufficiently small due to the divergence of the harmonic series . we therefore see that if @xmath966 and if @xmath626 is sufficiently small , then condition ( [ eq : convexity ] ) is satisfied but condition ( [ eq : tcondition ] ) is not . 99 j. baik , p. deift , and k. johansson , `` on the distribution of the length of the longest increasing subsequence of random permutations '' , _ _ , * 12 * , 11191178 , 1999 . j. baik , p. deift , and k. johansson , `` on the distribution of the length of the second row of a young diagram under plancherel measure '' , _ geom . and func _ , * 10 * , 2000 , 702731 . j. baik and e. m. rains , `` algebraic aspects of increasing subsequences '' , _ duke math . j. _ , * 109 * , 165 , 2001 . j. baik and e. m. rains , `` the asymptotics of monotone subsequences of involutions '' , _ duke math . j. _ , * 109 * , 205281 , 2001 . j. baik , p. deift , k. t .- mclaughlin , p. d. miller , and x. zhou , `` optimal tail estimates for directed last passage site percolation with geometric random variables '' , _ adv . _ , * 5 * , 12071250 , 2001 . j. baik , t. kriecherbauer , k. t .- mclaughlin , and p. d. miller , `` uniform asymptotics for polynomials orthogonal with respect to a general class of discrete weights and universality results for associated ensembles : announcement of results '' , _ internat . res . notices _ 15 , 821858 , 2003 . see also , by the same authors , `` uniform asymptotics for polynomials orthogonal with respect to a general class of discrete weights and universality results for associated ensembles '' , , submitted to _ ann . _ , 2003 . g. f. carrier , m. krook , and c. e. pearson , _ functions of a complex variable : theory and technique _ , hod books , ithaca , new york , 1983 . p. deift , `` integrable operators '' , in _ differential operators and spectral theory : m. sh . 70th anniversary collection _ ( v. buslaev , m. solomyak , d. yafaev , eds . ) , _ amer . math . _ , ser . 2 , * 159 * , amer soc . , providence , 1999 . a. deift , a. r. its and x. zhou , `` long - time asymptotics for integrable nonlinear wave equations '' , in _ important development in soliton theory _ , second edition ( a. s. fokas and v. e. zakharov , eds . ) , springer - verlag , berlin , 1993 . p. deift , t. kriecherbauer , k. t .- mclaughlin , s. venakides , and x. zhou , `` uniform asymptotics for polynomials orthogonal with respect to varying exponential weights and applications to universality questions in random matrix theory '' , _ comm . pure appl . math . _ , * 52 * , 13351425 , 1999 . p. deift , t. kriecherbauer , k. t .- mclaughlin , s. venakides , and x. zhou , `` strong asymptotics of orthogonal polynomials with respect to exponential weights '' , _ comm . pure appl . _ , * 52 * , 14911552 , 1999 . deift , s. venakides , and x. zhou , `` the collisionless shock region for the long - time behavior of solutions of the kdv equation '' , _ comm . pure appl . _ , * 47 * , 199206 , 1994 . p. deift , s. venakides , and x. zhou , `` new results in small dispersion kdv by an extension of the steepest descent method for riemann - hilbert problems '' , _ internat . res . notices _ 6 , 285299 , 1997 . p. deift and x. zhou , `` a steepest descent method for oscillatory riemann - hilbert problems : asymptotics for the mkdv equation '' , _ ann . _ , * 137 * , 295368 , 1993 . p. deift and x. zhou , `` asymptotics for the painlev ii equation '' , _ comm . pure appl . _ , * 48 * , 277337 , 1995 . p. a. deift and x. zhou , `` perturbation theory for infinite - dimensional integrable systems on the line . a case study '' , _ acta math . _ , * 188 * , 163262 , 2002 . a. fokas , a. its , and a. v. kitaev , `` discrete painlev equations and their appearance in quantum gravity '' , _ commun . _ , * 142 * , 313344 , 1991 . j. s. geronimo and k. m. case , `` scattering theory and polynomials orthogonal on the unit circle '' , _ j. math . _ , * 20 * , 299310 , 1979 . s. kamvissis , k. t .- mclaughlin , and p. d. miller , _ semiclassical soliton ensembles for the focusing nonlinear schrdinger equation _ , annals of math . studies , princeton university press , princeton , 2003 . mclaughlin and p. d. miller , in preparation . p. d. miller , `` asymptotics of semiclassical soliton ensembles : rigorous justification of the wkb approximation '' , _ internat . math . res . notices _ , no . 8 , 383454 , 2002 . n. i. muskhelishvili , _ singular integral equations _ , ( second edition ) , dover publications , new york , 1992 . p. g. nevai , `` orthogonal polynomials '' , _ mem . _ , * 18 * , 1979 . b. simon , _ orthogonal polynomials on the unit circle _ , to appear . g. szeg , _ orthogonal polynomials _ , ( fourth edition ) , american math . society colloquium publications , volume 23 , providence , 1975 . g. g. varzugin , `` asymptotics of oscillatory riemann - hilbert problems '' , _ j. math _ , * 37 * , 58695892 , 1996 .	we develop a new asymptotic method for the analysis of matrix riemann - hilbert problems . our method is a generalization of the steepest descent method first proposed by deift and zhou ; however our method systematically handles jump matrices that need not be analytic . the essential technique is to introduce nonanalytic extensions of certain functions appearing in the jump matrix , and to therefore convert the riemann - hilbert problem into a @xmath0 problem . we use our method to study several asymptotic problems of polynomials orthogonal with respect to a measure given on the unit circle , obtaining new detailed uniform convergence results , and for some classes of nonanalytic weights , complete information about the asymptotic behavior of the individual zeros .
the cumulative normal distribution , be it univariate or multivariate , has to be evaluated numerically . there are numerous algorithms available , many of these having been fine - tuned , leading to faster evaluation and higher accuracy but also to lack of mathematical transparency . for the univariate case , has proposed a very simple and intuitive but powerful alternative that is based on taylor expansion of mills ratio or similar functions . in this note we will extend marsaglia s approach to the bivariate case . this will require two steps : reduction of the evaluation of the cumulative bivariate normal distribution to evaluation(s ) of a univariate function , i.e. , to the cumulative bivariate normal distribution on the diagonal , and taylor expansion of that function . note that a similar approach , but with reduction to the axes instead of the diagonals , has been proposed by . the resulting algorithm has to be compared with existing approaches . for overview on and discussion of the latter , cf . @xcite , @xcite , @xcite , and @xcite . most implementations today will rely on variants of the approaches of or of . improvements of the latter method have been provided by and . the method of , although less reliable , is also very common , mainly because it is featured in @xcite and other prevalent books . it will turn out that the algorithm proposed in this paper is able to deliver near double precision ( in terms of absolute error ) using double arithmetic . furthermore , implementation of the algorithm using high - precision libraries is straightforward ; indeed , a quad - double implementation has been applied for testing purposes . performance is competitive , and trade - offs between speed and accuracy may be implemented with little effort . in this section we are going to develop the algorithm . in order to keep the presentation lean we will often refer to the author s recent survey @xcite . for further background on normal distributions the reader is also referred to text books such as @xcite , @xcite and @xcite . denote by @xmath0 the density and distribution function of the standard normal distribution . mills ratio is then defined as @xmath1 furthermore , denote by @xmath2 the density and distribution function of the bivariate standard normal distribution with correlation parameter @xmath3 . we will also write @xmath4 we are going to use the following properties : @xmath5 in the following we will assume that @xmath6 , @xmath7 . in this case the following bounds apply ( cf . 5.2 ) ) : @xmath8 furthermore , as is proven implicitly in ( * ? ? ? a.2 ) , @xmath9 now we define @xmath10 starting with @xmath11 we find the recursion @xmath12 which we can use to recursively evaluate the taylor expansion of @xmath13 around zero . dividing by @xmath14 for convenience , we define @xmath15 using @xmath16 we derive the following recursion scheme : @xmath17 with initial values @xmath18 here we have used that @xmath19 we can now compute @xmath20 numerically via @xmath21 note that it would also have been possible to work with , e.g. , one of the functions @xmath22 instead . the resulting recursion schemes are in fact easier ( two summands instead of three ) but will be running into numerical problems ( cancellation , or lower accuracy for @xmath23 ) . in order to apply the results from section [ subsec_diagonal ] to the numerical evaluation of @xmath24 for general @xmath25 , @xmath26 and @xmath27 , we start with the symmetric formula ( cf . * eq . ( 3.16 ) ) ) @xmath28 where @xmath29 and @xmath30 from the axis @xmath31 to the diagonal @xmath32 we get by applying the formula ( cf . * eq . ( 3.18 ) ) ) @xmath33 specifically , we obtain @xmath34 with @xmath35 where in an implementation ( [ eq_ax_rho1 ] ) should be used for @xmath36 , and ( [ eq_ax_rhom1 ] ) for @xmath37 , in order to avoid catastrophic cancellation . note also that @xmath38 in a last step , if necessary to ensure @xmath6 and @xmath7 , we apply the formulas ( cf . * eq . ( 2.15 ) ) and ( * ? ? ? * eq . ( 3.27 ) ) ) @xmath39 specifically , we obtain @xmath40 where in an implementation ( [ eq_bx_rho1 ] ) should be used for @xmath36 , and ( [ eq_bx_rhom1 ] ) for @xmath37 , in order to avoid catastrophic cancellation . it will be favorable to work with @xmath41 instead of @xmath42 . if ( [ eq_minusrho ] ) has to be applied ( i.e. , if @xmath43 , which is equivalent with @xmath44 ) , correspondingly we will work with @xmath45 in the following we will discuss implementation of the algorithm derived in section [ sec_theory ] . the c++ language has been chosen because it is the market standard in quantitative finance , one of the fields frequently requiring evaluation of normal distributions . source code ( in c++ ) for evaluation of @xmath20 as in section [ subsec_diagonal ] , for @xmath6 and @xmath7 , is provided in figure [ source_diag ] . in the following we will comment on some details of the implementation . equations ( [ eq_rec_1 ] ) - ( [ eq_start_6 ] ) show that it is reasonable to provide @xmath46 , instead of @xmath27 , as input for the evaluation of @xmath20 . moreover , cf . ( [ eq_rho_1 ] ) and ( [ eq_rho_2 ] ) , double inversion ( i.e. , computation of @xmath47 instead of @xmath48 ) is to be avoided in the reduction algorithm . values for @xmath49 and for @xmath50 are also expected as input parameters . this makes sense because the values are needed by the reduction algorithm as well ( and hence should not be computed twice ) . evaluation of @xmath51 is to be avoided for @xmath52 and has been replaced ( without optimization of the cutoff point ) by @xmath53 note that @xmath54 has to be computed anyway . constants ( all involving @xmath55 ) have been pre - computed in double precision . the recursion stops if a new term does not change the computed sum . if the a priori bound for the absolute error , given by ( [ eq_bounds ] ) , is less than @xmath56 , the upper bound is returned ( relative accuracy on the diagonal may be increased by dropping this condition but overall relative accuracy will still be determined by the reduction to the diagonal , cf . section [ subsec_reduction_imp ] ) , and by the accuracy of the implementation of @xmath57 . the final result is always checked against the upper and lower bound . note that @xmath58 and @xmath59 have different sign but comparable order . bracketing them before summation can therefore reduce cancellation error . ' '' '' .... double phi2diag ( const double & x , const double & a , // 1 - rho const double & px , // phi ( x ) const double & pxs ) // phi ( lambda ( rho ) * x ) { if ( a < = 0.0 ) return px ; // rho = = 1 if ( a > = 1.0 ) return px * px ; // rho = = 0 double b = 2.0 - a , sqrt_ab = sqrt ( a * b ) ; double asr = ( a > 0.1 ? asin ( 1.0 - a ) : acos ( sqrt_ab ) ) ; double comp = px * pxs ; if ( comp * ( 1.0 - a - 6.36619772367581343e-001 * asr ) < 5e-17 ) return b * comp ; double tmp = 1.25331413731550025 * x ; double a_coeff = a * x * x / b ; double a_even = -tmp * a ; double a_odd = -sqrt_ab * a_coeff ; double b_coeff = x * x ; double b_even = tmp * sqrt_ab ; double b_odd = sqrt_ab * b_coeff ; double d_coeff = 2.0 * x * x / b ; double d_even = ( 1.0 - a ) * 1.57079632679489662 - asr ; double d_odd = tmp * ( sqrt_ab - a ) ; double res = 0.0 , res_new = d_even + d_odd ; int k = 2 ; while ( res ! = res_new ) { d_even = ( a_odd + b_odd + d_coeff * d_even ) / k ; a_even * = a_coeff / k ; b_even * = b_coeff / k ; k++ ; a_odd * = a_coeff / k ; b_odd * = b_coeff / k ; d_odd = ( a_even + b_even + d_coeff * d_odd ) / k ; k++ ; res = res_new ; res_new + = d_even + d_odd ; } res * = exp ( -x * x / b ) * 1.591549430918953358e-001 ; return max ( ( 1.0 + 6.36619772367581343e-001 * asr ) * comp , b * comp - max ( 0.0 , res ) ) ; } .... ' '' '' source code ( in c++ ) for evaluation of @xmath20 as in equation ( [ eq_2axis ] ) is provided in figure [ source_phi ] , and source code for evaluation of @xmath60 is provided in figure [ source_help ] . in the following we will comment on some details of the implementation . the special cases @xmath61 and @xmath62 are dealt with in phi2 ( ) . therefore , in phi2help ( ) there is no check against 1.0 - rho = = 0.0 , 1.0 + rho = = 0.0 or s = = 0.0 . it is assumed that sqr(x ) evaluates xx . the cutoff points @xmath63 have been set by visual inspection and might be optimized . ' '' '' .... double phi2help ( const double & x , const double & y , const double & rho ) { if ( x = = 0.0 ) return ( y > = 0.0 ? 0.0 : 0.5 ) ; double s = sqrt ( ( 1.0 - rho ) ( 1.0 + rho ) ) ; double a = 0.0 , b1 = -fabs ( x ) , b2 = 0.0 ; if ( rho > 0.99 ) { double tmp = sqrt ( ( 1.0 - rho ) / ( 1.0 + rho ) ) ; b2 = -fabs ( ( x - y ) / s - x * tmp ) ; a = sqr ( ( x - y ) / x / s - tmp ) ; } else if ( rho < -0.99 ) { double tmp = sqrt ( ( 1.0 + rho ) / ( 1.0 - rho ) ) ; b2 = -fabs ( ( x + y ) / s - x * tmp ) ; a = sqr ( ( x + y ) / x / s - tmp ) ; } else { b2 = -fabs ( rho * x - y ) / s ; a = sqr ( b2 / x ) ; } double p1 = phi ( b1 ) , p2 = phi ( b2 ) ; // cum . standard normal double q = 0.0 ; if ( a < = 1.0 ) q = 0.5 * phi2diag ( b1 , 2.0 * a / ( 1.0 + a ) , p1 , p2 ) ; else q = p1 * p2 - 0.5 * phi2diag ( b2 , 2.0 / ( 1.0 + a ) , p2 , p1 ) ; int c1 = ( y / x > = rho ) ; int c2 = ( x < 0.0 ) ; int c3 = c2 & & ( y > = 0.0 ) ; return ( c1 & & c3 ? q - 0.5 : c1 & & c2 ? q : c1 ? 0.5 - p1 + q : c3 ? p1 - q - 0.5 : c2 ? p1 - q : 0.5 - q ) ; } .... ' '' '' ' '' '' .... double phi2 ( const double & x , const double & y , const double & rho ) { if ( ( 1.0 - rho ) * ( 1.0 + rho ) < = 0.0 ) // \|rho\| = = 1 if ( rho > 0.0 ) return phi ( min ( x , y ) ) ; else return max ( 0.0 , min ( 1.0 , phi ( x ) + phi ( y ) - 1.0 ) ) ; if ( x = = 0.0 & & y = = 0.0 ) if ( rho > 0.0 ) return phi2diag ( 0.0 , 1.0 - rho , 0.5 , 0.5 ) ; else return 0.5 - phi2diag ( 0.0 , 1.0 + rho , 0.5 , 0.5 ) ; return max ( 0.0 , min ( 1.0 , phi2help ( x , y , rho ) + phi2help ( y , x , rho ) ) ) ; } .... ' '' '' evaluation of @xmath24 as in section [ sec_implement ] will require ( at most ) four calls to an implementation of the cumulative standard normal distribution @xmath57 ( phi ( ) in the code ) . the actual choice may well determine both accuracy and running time of the algorithm . for testing purposes i have been using a hybrid method , calling the algorithm from ( * ? ? ? 2 ) for absolute value larger than @xmath64 , and phi ( ) from @xcite else . besides phi ( ) , exp ( ) will be called two times , arcsin ( ) or arccos ( ) two times , and sqrt ( ) six times . everything else is elementary arithmetic . due to the reduction algorithm , the final result will be a sum . therefore , very high accuracy in terms of relative error can not be expected . consequently , evaluation of the diagonal aims at absolute error as well . the phi2diag ( ) function is behaving as it may be expected from an approximation by a taylor series around zero : ( absolute ) error increases with decreasing @xmath25 . for @xmath65 ( or @xmath66 or @xmath36 ) the error bounds from ( [ eq_bounds ] ) are taking over , and absolute error decreases again . the maximum absolute error is obtained for @xmath67 , @xmath68 ( maximum error of the upper bound is obtained for @xmath69 , cf . 5.2 ) ) . in general , assuming that all numerical fallacies in the reduction algorithm have been taken care of , the diagonal is expected to provide a worst case because the errors of the two calls to phi2diag ( ) will not cancel . with respect to the reduction algorithm , the case @xmath70 , @xmath71 , implying @xmath72 , is most critical . in order to give an impression of the algorithm s behaviour , we will discuss the results of a simulation study . for each @xmath73 , @xmath74 , the value of @xmath75 has been computed via the phi2 ( ) function from figure [ source_phi ] where @xmath76 has been drawn from a uniform distribution on @xmath77 $ ] with @xmath78 , @xmath79 has been drawn from a uniform distribution on @xmath80 $ ] , and @xmath81 where @xmath82 has been drawn from a uniform distribution on @xmath80 $ ] as well . the c++ implementation from @xcite has been serving as a competitor . both functions have been evaluated against a quad - double precision version of phi2 ( ) , implemented using the qd library @xcite and quad - double precision constants . the diagram in figure [ diag_error ] is displaying , for @xmath73 , the 99% quantile and the maximum of the absolute difference between the double precision algorithms ( phi2 and west ) and the quad - double precision algorithm . apart from a shift due to subtractions for positive @xmath25 , errors of phi2 are rather symmetric around zero . the peaks at @xmath83 are due to the taylor expansion around zero ; the peaks at @xmath84 are due to taylor expansion after transformation of the argument . the characteristics of the @xmath85 quantile , in particular the little peaks at @xmath86 , are already visible in the error of the @xmath57 function used . the maximum error of west almost always stays below the one of phi2 . note that the maximum error of west is determined by the case @xmath37 and might be reduced by careful consideration of that case . in the simulation study , phi2 was a little slower than west : it took approximately five minutes and four minutes to perform the @xmath87 evaluations on a fairly standard office pc ( and it took two days to perform the corresponding quad - double precision evaluations ) . the number of recursion steps used by phi2diag is increasing with @xmath88 . because of the mathematical transparency of the algorithm it should be easy to find an appropriate trade - off between speed and accuracy by replacing the condition terminating the recursion . wang , m. , kennedy , w.j . ( 1990 ) , _ comparison of algorithms for bivariate normal probability over a rectangle based on self - validated results from interval analysis _ , journal of statistical computation and simulation 37(1 - 2 ) , pp .	we propose an algorithm for evaluation of the cumulative bivariate normal distribution , building upon marsaglia s ideas for evaluation of the cumulative univariate normal distribution . the algorithm is mathematically transparent , delivers competitive performance and can easily be extended to arbitrary precision .
the existence of ultrahigh energy ( uhe ) cosmic rays comprised of nucleons guarantees the presence of ultrahigh energy neutrinos . over megaparsec distances , uhe nucleons interact with the 2.7k microwave background radiation to produce the delta resonance , which decays to a pion and a nucleon . charged pions decay to neutrinos and muons , followed by muon decay , leading to fluxes of electron and muon neutrinos . beginning with the pioneering work of ref . 1 , and continuing to recent improvements@xcite , there are quantitative predictions for fluxes of neutrinos with @xmath3 gev and higher energies . at somewhat lower energies , one expects that sources of tev photons such as active galactic nuclei should also produce neutrinos . the attenuation length of the photon effectively cuts off the photon energy at @xmath4 tev for sources farther than 100 mpc from the earth . the long interaction length of neutrinos means that neutrino attenuation from distant sources is not an issue over astronomical distances due to low densities . a common feature to neutrino flux models is that roughly two muon neutrinos / antineutrinos appear for each electron neutrino / antineutrino at the production site . ( in the following discussion , we do not distinguish between @xmath5 and @xmath6 . ) in the past few years , the superkamiokande experiment@xcite together with the sno@xcite results point to physics beyond the standard model : neutrino masses and mixing . parameters in the mixing matrix lead fairly robustly to an equal distribution of fluxes of neutrinos between the three neutrino flavors@xcite : @xmath7 after propagating over astronomically large distances . uhe tau neutrino interactions with nucleons in the earth and atmosphere may provide signals of this phenomenon . tests for deviations from the standard model ultrahigh energy cross section are another probe of non - standard model physics with uhe neutrinos . the neutrino energy regime determines the detection method . for @xmath8 gev , underground detectors can be used to record upward muons and electromagnetic / hadronic showers . experiments include water or ice cherenkov detectors like amanda@xcite , nestor@xcite , antares@xcite and icecube@xcite . attenuation of the neutrino flux in earth is important , especially the difference between @xmath9 and @xmath10 attenuation@xcite . in the energy range above @xmath11 gev , detection of horizontal air showers and radio detection are used to determine the uhe @xmath5 fluxes and to probe the neutrino cross section@xcite . at intermediate energies , the double - bang signal of both production and decay vertices of the @xmath12 process can be used to identify @xmath10s@xcite . other detection possibilities include using the earth as a tau neutrino converter , then searching for the shower from the decay of the emerging tau@xcite . we present results in two of these examples : upward neutrino fluxes and their detection , and the potential for limits on the non - standard neutrino cross section from the owl@xcite proposed detector . in this latter case , the non - standard model considered is mini - black hole production and decay@xcite , a feature of models with tev - scale compactification of extra dimensions@xcite . before going into detail of these two examples , we review the theoretical status of the neutrino cross section at high energies the neutrino - nucleon cross section is straightforward to calculate in the standard model@xcite . the charged current cross section is @xmath13\ ] ] for @xmath14 boson momentum squared @xmath15 , lepton energy difference @xmath16 , bjorken @xmath17 and @xmath18 with nucleon mass @xmath19 and @xmath14 mass @xmath20 . the parton distribution function combinations @xmath21 and @xmath22 are detailed in , for example , ref . the cross section depends crucially on the parton distribution functions@xcite . as @xmath23 gets large , the parton distributions at small-@xmath24 increase , however , the @xmath14 boson propagator decreases with increasing @xmath25 . at sufficiently high energies , the value of @xmath25 saturates at @xmath26 , so the relevant values of @xmath24 for a given incident neutrino energy @xmath27 is approximately @xmath28 this translates to @xmath29 for @xmath30 gev , and @xmath31 for @xmath32 gev . measurements of the parton distribution functions have been made down to @xmath33 at @xmath34 gev@xmath35 at hera@xcite . at a more relevant @xmath26 , parton distributions have been measured to @xmath36@xcite . beyond the measured regime , one is guided by theoretical arguments concerning the small-@xmath24 parton distribution functions and their qcd evolution . as discussed in detail in ref . 35 , for @xmath37 at @xmath38 , then @xmath39 for @xmath40 . at small values of @xmath24 , the quark distribution functions are dominated by the quark sea , which come from @xmath41 , so the sea distributions have the same power law dependence : @xmath42 . modern parton distribution function fits have @xmath43 , and extrapolations to small-@xmath24 by this power law agree well with perturbative qcd evolution , as demonstrated , for example , by glck , kretzer and reya in ref . kwiecinski , martin and stasto have used a bfkl - type evolution at small-@xmath24 and obtain cross sections that are in agreement with power law extrapolations@xcite . at very small values of @xmath24 , recombination processes such as @xmath44 should become important , ultimately leading to a flattening of the neutrino cross section as a function of energy . the recombination regime is unimportant for neutrino energies up to @xmath0 gev , though it should become important at even higher energies@xcite . 1 shows the neutrino interaction length , proportional to the inverse of the cross section : @xmath45 with @xmath46 equal to avogadro s number . for neutrino energies above a few tens of tev , the earth s diameter is larger than the attenuation length . even for neutrinos with @xmath47 gev , the horizontal atmospheric depth is short compared to the neutrino interaction length . in our discussion of upward neutrinos , we restrict our attention to neutrino energies below 10@xmath48 gev@xcite . for the upward neutrino flux , depending on the nadir angle , attenuation due to neutrino passage through the earth is important . neutrino oscillations of @xmath49 lead to the arrival of equal fluxes of @xmath9 and @xmath10 at the surface of the earth , however , it has been noted that the attenuation of the two species is quite different due to the short lifetime of the @xmath50@xcite . muon and tau neutrinos have the same neutral current and charged current cross sections above 1 tev . ( at 1 tev , the charged current cross section for @xmath51 is about 5% lower than the @xmath9 charged current ( cc ) cross section due to the @xmath50 mass . at 100 gev , the @xmath10 cc cross section is about 25% smaller@xcite . ) the difference in the attenuation of @xmath9 and @xmath10 fluxes has to do with the fact that muons lose energy as they traverse a medium before they decay . electromagnetic energy loss for muons is significant , so the neutrinos from muon decay have much lower energies than the parent muon energy . on the other hand , electromagnetic energy loss for @xmath50 s is important only for tau energies above @xmath52 gev@xcite . the muon neutrino and tau neutrino transport equations are @xmath53 \\ \nonumber & & { \partial f_{{\tau}}(e , x)\over \partial x}\simeq -{f_{{\tau}}(e , x)\over { \cal l}_\tau^{dec } } \\ \nonumber & & \quad\quad + \int_e^\infty de_y\,g^{\nu_\tau\rightarrow \tau}(e , e_y , x)\ ] ] where , for example , @xmath54{dn^{nc } \over de}(e_y , e)\ .\ ] ] in terms of the neutrino flux @xmath55 and the cross section normalized neutral current energy distribution @xmath56 . we have made the approximation that tau energy loss is negligible , and that only the decay of the tau is relevant , neglecting tau cc interactions . the feed - down of neutrinos from @xmath51 is evident in comparison of the attenuated @xmath10 flux with the @xmath9 flux@xcite . in figs . 2 and 3 we show the ratio of the attenuated flux to the incident flux for @xmath10 ( upper curve ) and @xmath9 ( lower curve ) at nadir angle 0 for two different fluxes , parameterized by @xmath57 for @xmath58 gev . 2 , @xmath59 and @xmath60 , and in fig . 3 , @xmath61 and @xmath62 . the tau neutrino pile - up is more pronounced for the flatter fluxes . at larger nadir angles , the pileup reduces because the number of interactions reduces , but the attenuation is also less . experimentally , one would like to exploit the effect of the tau neutrino pile - up and to identify the tau neutrino component of the flux . in a manner similar to the lower energy comparison of the neutral current to charged current event rates@xcite , for neutrinos with energies above 1 tev , one can compare the upward electromagnetic / hadronic shower event rate from @xmath63 with the upward muon event rate from @xmath64 we have made a detailed comparison in ref . 12 . by measuring the upward shower and muon event rates with an energy threshold of 10 tev , one should be able to see the effect of the tau neutrino component of the incident neutrino flux . this is relatively independent of the _ a priori _ unknown spectrum of the incident flux , except for the overall normalization ( so that there are enough events ) . an interesting possibility for a class of input neutrino fluxes is the enhancement of the upward muon rate from @xmath65 where the final muon neutrino interacts to produce a muon@xcite . the muon neutrinos from tau decays are called secondary neutrinos . " there is incremental energy loss at each of the processes : cc production of the tau , then muonic decay of the tau . this results in a contribution to the muon neutrino flux@xcite , as seen in fig . 4 for @xmath66 for nadir angle 0 . the upper dashed line is @xmath67 for the @xmath10 flux , the lower dashed line is for @xmath9 excluding the secondary neutrinos , while the solid line is the secondary muon neutrino flux . the implications for the upward muon rate are shown in fig . 5 . the secondary neutrino flux for the @xmath68 incident flux is relatively less important . for less steep fluxes than @xmath69 , one expects that the secondary flux will be enhanced . as fig . 1 indicates , the standard model neutrino interaction length is large compared to the horizontal atmospheric depth ( zenith angle @xmath70 at sea level ) . unlike strongly interacting particles , neutrinos can penetrate deep into the atmosphere . horizontal air showers initiated deep in the atmosphere are a signal of neutrinos . non - observation of an enhancement of horizontal air shower event rates limit a combination of the neutrino - nucleon cross section and the neutrino flux@xcite . using the cosmogenic neutrino flux from cosmic ray scattering with the 2.7k photons , it is possible to set limits on non - standard contributions to the cross section . the cosmogenic neutrino flux , for the sum of electron neutrino plus antineutrino , evaluated by engel , seckel and stanev@xcite ( ess ) is shown in fig . their results are shown for two models of source evolution : one parameterized like @xmath71 for redshift @xmath72 ( standard evolution ) , and the other scaling like @xmath73 for the same redshift ( strong evolution ) . discussed here are contributions from mini - black holes , predicted by models with the planck scale in the tev range and a number of extra dimensions@xcite , and the potential of an orbiting observatory to detect the enhanced rate of horizontal air showers@xcite . there is an extensive literature on black hole production and detection by neutrino telescopes@xcite . it appears that enhancements of the cross section from @xmath74-channel kaluza - klein graviton exchange is not an important effect@xcite . black hole production in these models is determined by the geometrical cross section , depending on the schwarzschild radius . the schwarzschild radius @xmath75 , assuming @xmath76 compactified dimensions and planck scale @xmath77 , is given by@xcite @xmath78^{\frac{1}{1 + n}}\ .\ ] ] here , @xmath79 is the mass of the black hole . the geometrical cross section for neutrino interaction with parton @xmath80 is then @xmath81 for @xmath82 equal to the neutrino - parton center of mass energy . ( 10 ) enforces the requirement that @xmath82 be larger than some minimum energy - squared , certainly larger than the planck scale . convoluting with the parton distribution functions @xmath83 after setting @xmath84 give a non - standard model contribution to the neutrino - nucleon cross section : @xmath85 the black hole is expected to rapidly evaporate into standard model particles , and we assume that all of the energy of the black hole goes into the shower it produces@xcite . experimental constraints on deviations from newtonian gravity exclude @xmath86 and @xmath87 tev@xcite . for @xmath88 , @xmath89 tev is consistent with tests of the inverse square law,@xcite while for larger @xmath76 , there are no constraints from tests of newtonian gravity for @xmath87 tev . collider experiments put a lower limit on @xmath77 at approximately 1 tev@xcite . 7 shows a representative set of parameter choices for the number of extra dimensions @xmath76 and the minimum black hole mass , given a planck scale of @xmath90 tev . also shown are curves representing upper limits on the cross section from tyler , olinto and sigl@xcite ( tos ) and anchordoqui , feng , goldberg and shapere@xcite ( afgs ) , were they have used cosmogenic fluxes similar to that of ess . these upper limits come from the agasa and fly s eye limits on an excess of horizontal air shower events . the largest enhancement of the cross section comes at the highest energies . detectors optimized to this energy range will be able to put the most stringent limits on the cross section . the proposed orbiting wide - angle light - collectors experiment@xcite ( owl ) and the extreme universe space observatory@xcite ( euso ) are two such detectors . the owl experiment would involve two satellites with photodetectors , orbiting 640 km above the earth . the euso experiment would be located on the international space station , 380 km above the earth . the conversion factor from the 2 telescopes to one at a reduced altitude amount to a reduction of the event rate by a factor of about 0.2 compared to owl . the threshold for owl detection of horizontal air showers is on the order of @xmath2 gev . by rescaling the effective aperture for owl detection of electron neutrinos by the ratio of the standard model plus black hole cross section to the standard model neutrino - nucleon cross section , one can arrive at event rate predictions given the ess incident electron neutrino flux . the range of electron neutrino induced event rates goes from 100 events per year for @xmath91 and @xmath92 tev to less than one event per year for @xmath93 tev and @xmath94 for the standard evolution of the cosmogenic flux as shown in fig . the standard model event rate is about 0.5 events / year . the strong evolution of the cosmogenic flux gives a factor of 2 enhancement for both signal and background . the owl rates are about a factor of 20 times larger than the icecube downward contained event rates for an energy threshold of @xmath95 gev . icecube will optimally detect fluxes with large low energy components , _ e.g. _ , @xmath96@xcite . if all three neutrino flavors are included in the event rate calculation , the standard model rates increase by a factor of about 2 , while the black hole rates are increased by a factor of 3 since the black hole cross section is flavor - blind . ( the standard model rate depends on how much of the incident neutrino energy is translated to the shower in the final state . ) including these factors , one finds that owl would be able to probe the fundamental planck scale up to 3 tev for @xmath97 even with one year of data taking . described above are two ways to search for nonstandard model physics with ultrahigh energy neutrinos : searching for evidence of @xmath98 oscillations with tau appearance in upward muon and upward shower rates , and mini - black hole production in neutrino induced horizontal air showers . nonstandard model physics may be manifest in the flux of neutrinos , through exotic particle decay to neutrinos@xcite . a wide range of experiments : underground cherenkov detectors , surface arrays and satellite light collectors cover a huge range of energies from gev to 10@xmath99 gev incident neutrinos . the view from the ultrahigh energy neutrino window on astrophysics and particle physics should provide interesting tests of the standard model and its extensions . the work of s.i.d . has been supported in part by national science foundation grant no . 0070998 . the work of i.s . has been supported in part by the doe under contracts no . de - fg02 - 95er40906 and de - fg02 - 93er40792 . the work of m.h.r . has been supported in part by the doe under contract no . fg02 - 91er40664 . k. greisen , phys . * 16 * , 748 ( 1966 ) ; v. berezinsky and g. zatsepin , in _ proceedings of the 1976 dumand summer workshop _ , ed . a. roberts and r. donaldson ( fermilab , batavia , illinois , 1977 ) p. 215 . c.t . hill and d.n . schramm , phys . d * 31 * , 564 ( 1985 ) ; s. yoshida and m. teshima , prog . . phys . * 89 * , 833 ( 1993 ) ; r.j . protheroe and p.a . johnson , astropart . phys . * 4 * , 253 ( 1996 ) ; * 5 * , 215(e ) ( 1996 ) . [ owl / airwatch collaboration ] , in _ proceedings of the 26th international cosmic ray conference ( icrc 99 ) _ ( 1999 ) vol . 2 , 388 - 391 ; http://lheawww.gsfc.nasa.gov/docs/gamcosray/hecr/ owl_new.html . j. l. feng and a. d. shapere , phys . rev . lett . * 88 * , 021303 ( 2002 ) ; l. anchordoqui and h. goldberg , phys . rev . d * 65 * , 047502 ( 2002 ) ; y. uehara , prog . . phys . * 107 * , 621 ( 2002 ) . a. ringwald and h. tu , phys . b * 525 * , 135 ( 2002 ) ; m. kowalski , a. ringwald and h. tu , phys . lett . * b529 * , 1 ( 2002 ) . i. antoniadis , phys . b * 246 * , 377 ( 1990 ) ; n. arkani - hamed , s. dimopoulos and g. r. dvali , phys . b * 429 * , 263 ( 1998 ) ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys . b * 436 * , 257 ( 1998 ) ; l.j . randall and r. sundrum , phys . lett . * 83 * , 3370 ( 1999 ) . see , for example , c. quigg , _ gauge theories of the strong , weak , and electromagnetic interactions _ ( benjamin / cummings , menlo park , 1983 ) . s. b. giddings and s. thomas , phys . d * 65 * , 056010 ( 2002 ) ; s. dimopoulos and g. landsberg , phys . lett . * 87 * , 161602 ( 2001 ) ; t. g. rizzo , in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed . r. davidson and c. quigg , hep - ph/0111230 ; k. cheung , phys . d * 66 * , 036007 ( 2002 ) .	the ultrahigh energy neutrino cross section is well understood in the standard model for neutrino energies up to @xmath0 gev . tests of neutrino oscillations @xmath1 from extragalactic sources of neutrinos are possible with large underground detectors . measurements of horizontal air shower event rates at neutrino energies above @xmath2 gev will be able to constrain nonstandard model contributions to the neutrino - nucleon cross section , _ e.g. _ , from mini - black hole production .
arp 299 is a nearby luminous infrared galaxy ( d@xmath643 mpc for h@xmath7=70 km s@xmath8mpc@xmath8 , @xmath9= 0.7 , @xmath10= 0.3 ) with an infrared ( 8 - 1000 @xmath11 m ) luminosity of l@xmath12=5.7@xmath210@xmath13l@xmath14 derived from the equation detailed on sanders & mirabel ( 1996 ) . ] . it is an interacting system ( fig . [ galaxia ] ) composed of two individual galaxies ( ic 694+ngc 3690 ( see also alonso - herrero et al 2000 ) . ] ) in an early dynamical stage . hibbard & yun ( 1999 ) have discovered an hi tidal tail of 180 kpc in length , with no evidence for tidal dwarf galaxies . powerful starburst regions with star formation rates of up to about 100 m@xmath14 yr@xmath8 have been identified ( alonso - herrero et al . 2000 , hereafter aah00 ) . moreover , the system presents several regions where supernova remnants have been detected at radio wavelengths ( neff et al . more recently , mattila et al . ( 2005 ) have discovered a new supernova using near - ir images , reinforcing the prediction that this galaxy has a high rate of supernova explosions ( mannucci et al . 2003 ) . optical spectroscopic studies have classified ic 694 as starburst , and ngc 3690 as starburst / liner ( e.g. , coziol et al . 1998 ) , and mid - ir observations classified the system as starburst ( e.g. , laurent et al . 2000 ) . using _ hubble space telescope _ ( hst ) imaging and near - infrared spectroscopy , aah00 modeled the star formation properties of the system without the need for an agn . the first hint for the presence of an agn in arp 299 came from x - ray observations with the _ bepposax _ satellite by della ceca et al . these authors concluded that a highly absorbed ( n@xmath15 @xmath16 2.5@xmath210@xmath17 @xmath18 ) agn , with an intrinsic luminosity of l@xmath19 @xmath16 1.9@xmath210@xmath20 erg s@xmath8 , is present in this galaxy , although due to the low spatial resolution its exact location could not be determined . using _ chandra _ and _ xmm - newton _ observations , recent studies ( ballo et al . 2004 ) have inferred the presence of two agns , one located in ngc 3690 ( source b1 ; 6.4 kev line ) , and another in ic 694 ( source a ; 6.7 kev fe - h@xmath1 line ) . in addition , gallais et al . ( 2004 ) suggest the presence of an agn in ngc 3690 based on the shape of the mid - ir continuum , while in ic 694 they only detect spectral features characteristic of massive , dust enshrouded star formation . this paper presents a detailed study of the two - dimensional star formation , kinematics , and ionized gas properties of arp 299 , covering the two individual galaxies of the system and the interface area . at a distance of about 43 mpc , arp 299 is one of the nearest interacting , luminous infrared galaxies , and therefore these observations provide high spatial resolution ( about 0.205 kpc arcsec@xmath8 ) and allow us to investigate in detail the structure and physical properties of this galaxy . the paper is organized as follows : 2 describes the observations and data reduction , while the selection of regions under study is described in 3 . the next sections present a discussion of the results , with special emphasis on the stellar and ionized gas morphology ( [ sec4 ] ) , the internal dust distribution and extinction effects ( [ sec5 ] ) , the extinction estimates using optical and near - ir emission lines ( [ sec54 ] ) , the two dimensional ionization structure ( [ sec6 ] ) , the electron density ( [ electron ] ) , the origin of the liner - like ionization ( [ sec8 ] ) , the star - forming properties ( [ sec9 ] ) , and the kinematics and dynamical mass derivations ( [ sec - kin ] ) . finally [ sec11 ] presents a short summary of the main new results . integral field spectroscopy ( ifs ) of arp 299 was obtained on 2004 january 15 and 17 using integral , a fiber - based integral field system ( arribas et al . 1998 ) connected to the wide - field fiber optic spectrograph ( wyffos ) ( bingham et al . 1994 ) and mounted on the 4.2 m william herschel telescope . two different integral configurations were used for observing the system ( see pointings in fig . [ galaxia ] ) . the fiber bundle with the largest field of view ( fov ) , sb3 , was centered on a position equidistant from the centers of the galaxies and therefore used to map the diffuse , interface region between the two galaxies ( ic 694+ngc 3690 ) that form the arp 299 system , as well as the galaxy external regions at distances of about 4 kpc . the sb3 bundle consists of 115 fibers each 27 in diameter , covering a rectangular area of 336@xmath2294 on the sky . in addition 20 fibers , forming a circle of 450 in radius , were used to obtain a spectrum of the sky simultaneously . the sb2 bundle was used to map each of the two galaxies ( ic 694 and ngc 3690 , eastern and western galaxies , respectively on fig . [ galaxia ] ) individually . it consists of 189 fibers each 09 in diameter , covering a rectangular area of 160@xmath2123 on the sky . in addition 30 fibers , forming a circle of 450 in radius , were used to measure the sky simultaneously . the spectra were taken using a 600 lines mm@xmath8 grating , with an effective spectral resolution ( fwhm ) of approximately 6.0 and 9.8 for the sb2 and sb3 bundles respectively , and covering the spectral range @xmath214400@xmath227000 . the total integration time was 3600 s for each pointing with the sb2 bundle , split into three individual exposures of 1200 s each . the total integration time with the sb3 bundle was 5400 s divided into three individual exposures of 1800 s each . the seeing conditions were fairly stable during the two periods of observation ( about 07 the first night , when we used the sb2 bundle , and 11 the second night when we used the sb3 bundle ) . the data reduction consists of three main steps : 1 ) the basic reduction of two dimensional fiber spectra , 2 ) the line fitting , and 3 ) the generation of maps of spectral features ( such as intensity maps and velocity fields ) from the calibrated spectra . the basic reduction was performed using the iraf environment , and followed the standard procedures applied to spectra obtained with two - dimensional fiber spectrographs ( arribas et al . 1997 and references therein ) . the absolute flux calibration was performed observing the standard star feige 34 with the same instrument configuration and data reduction procedures used on the galaxy . for each spectrum , the total flux , the radial velocities and the velocity dispersions ( @xmath23 ) were measured by fitting single gaussian functions to the observed emission line profiles using the dipso package ( howarth & murray 1988 ) . for the generation of the maps , a two dimensional interpolation was applied using the free software ida ( garca - lorenzo et al . the blue and red stellar continuum images were taken as an emission line - free filter centered at @xmath24=4475 and 6525 respectively , and with a rectangular bandwidth of 150 . the measures on the sb3 maps were done through an aperture of 30 in diameter , while for the sb2 maps the apertures diameter were 10 ( ic 694 ) and 09 ( ngc 3690 , to avoid the overlap between regions ) . three sample spectra , covering the entire used spectral range , are shown in fig . [ espectrillos ] . all the spectra show the strong ( h@xmath25 , [ o iii]@xmath214959 , 5007 , h@xmath1+[nii]@xmath216548 , 6584 , hei@xmath06678 , and [ sii]@xmath216716 , 6731 ) and the weak ( ni@xmath05199 , hei@xmath05876 , [ oi]@xmath216300 , 6364 , and hei@xmath06678 ) emission lines in different ratios , as well as interstellar absorption lines ( nai@xmath215890 , 5896 ) , that will be used to trace the neutral interstellar medium kinematics . complementary hst multiwavelength images both in continuum and emission lines are also available . these additional data include the red ( wfpc2-f814w , archive images ) and the near - infrared ( nicmos1-f110 m , nicmos2-f160w and nicmos2-f222 m , from aah00 ) continuum images . these filters are the hst equivalent to the ground - based johnson - cousins system _ i _ , _ j _ , _ h _ and _ k _ respectively ( origlia & leitherer 2000 ) . images of the ionized gas as traced by the pa@xmath1 ( @xmath26=1.875 @xmath11 m ) light ( nicmos - f190n , from aah00 ) will be used too . the archival wfpc2 image was calibrated on - the - fly with the best available reference files , while the nicmos images were reduced with routines from the package nicred ( mcleod 1997 ; see details on aah00 ) . in order to analyze the system we have selected several regions distributed over the individual galaxies , the interface between them and regions external to the galaxies ( fig . [ galaxia ] ) . the selection includes compact unresolved high - surface brightness knots , star clusters and more diffuse , extended resolved low - surface brightness areas . for ic 694 and ngc 3690 we based our selection on that done by aah00 , that includes the nuclear regions ( a and b1 respectively ) and the most important star - forming regions and bright star clusters . to identify the position of the regions under study in the lower resolution integral maps , a direct comparison of the convolved wfpc2 image with the red continuum image was performed . the following sections present the main results obtained for these different regions and discuss their implications separately for the two individual galaxies , ic 694 and ngc 3690 , and for the interface region between them . the physical properties measured for each individual region are presented in tables 1 to 3 . arp 299 is the result of a merger between the galaxies ic 694 and ngc 3690 , whose nuclei ( a and b1 respectively , see fig . [ interfaceinfo ] ) are located about 4.6 kpc apart . the system shows a morphologically disturbed structure characterized by the presence of compact high - surface knots , and a complex system of filaments and dust lanes , as already indicated by previous studies at several wavelengths ( e.g. gerhz et al . 1983 ; telesco et al . 1985 ; baan et al . 1990 ; aah00 ) . in what follows , we discuss the differences in the spatial distribution of the stellar and ionized gas components . the two integral / sb3 continuum images of arp 299 , obtained at wavelengths blueward of h@xmath1 and h@xmath25 , reproduce well the overall stellar morphology detected with the hst wfpc2 f814w image ( fig . [ interfaceinfo ] ) although at a lower spatial resolution ( fiber diameter 27 , scales of @xmath160.5 kpc ) . in spite of that , the interface region between the two galaxies , and the external regions show well defined features such as the bridge to the south of region k5 connecting ngc 3690 and ic 694 , and the elongated emission connecting k1 towards k12 source as part of a tail - like structure located to the sw of ngc 3690 ( see fig . [ galaxia ] ) . the general structure of the low surface brightness ionized gas in the interface region , as traced by for example h@xmath1 ( fig . [ interfaceinfo ] ) , is similar to that of the stellar component . however , several differences exist . diffuse ionized gas is detected ne of ngc 3690 , associated with the k5 and c@xmath27 knots ; the latter region shows pa@xmath1 emission and sn activity in radio ( neff et al . 2004 ) , and is part of a bridge of ionized material ( see fig . [ interfaceinfo ] , h@xmath1 map ) to the north of the stellar bridge that connects the two individual galaxies . the [ o]@xmath05007 emission map presents an elongated structure located towards the south - east of b1 and b2 tracing an area where the ionized gas presents high excitation state ( see [ ionizinterface ] ) . however , as showed in the next two sections , the most important morphological variations are associated to the main bodies of the galaxies when observed with the higher angular resolution sb2 bundle . previous studies have concluded that ic 694 was a spiral galaxy that due to the merging process shows a more disrupted morphology ( e.g. , gerhz et al . 1983 ; standford & wood 1989 ; hibbard & yun 1999 ) . the stellar structure as seen from both the hst and the integral continuum images ( see fig . [ icinfo ] ) is divided into two main substructures : ( 1 ) the arm ( region centered around a7 ) of about 1.8 kpc in length , that could be one of the original spiral arms disrupted and deformed as a consequence of the merging - process , and ( 2 ) the main body of the galaxy , formed by the dust - enshrouded infrared nucleus a ( aah00 ) , and surrounding regions . the two structures are well separated ( distances of @xmath161 kpc ) by what seems to be a dust lane , that is clearly visible in the hst near - ir color maps ( aah00 ) . the effect of the internal extinction is remarkable in the continuum images as the relative surface brightness of the regions , in particular nucleus a and surrounding regions , changes with wavelength ( see fig . [ icinfo ] and discussion in [ sec52 ] ) . the ionized gas distribution ( see fig . [ icinfo ] ) shows several distinct features worth mentioning . the nucleus of the galaxy ( a ) appears as a weak optical line emitting source , whereas the regions of high surface brightness are located at different positions ( e.g. , part of the arm and towards east , and regions close to a ) and separated up to 1 kpc from the nucleus . however , in the near - ir a is the brightest pa@xmath1 and br@xmath28 source ( see aah00 and sugai et al . 1999 , see also [ sec52 ] ) , indicating that the extinction towards the nucleus is very high ( using pa@xmath1/h@xmath1 ratio , a@[email protected] ) . there are regions ( e.g. knot g ) that appear as a faint continuum sources , but that are clearly identified as strong optical line emitters . in addition to differences in the morphology of the ionized gas ( i.e. excitation conditions ) with respect to that of the continuum , there are also differences in the ionized gas distribution as traced by several lines . the hydrogen recombination lines and the [ nii]@xmath06584 and [ sii]@xmath216717 , 6731 emission lines show a similar distribution , dominated by the emission from regions located on the arm and towards the east , and in the regions surrounding the nucleus . however , the shock tracer [ oi]@xmath06300 line is concentrated in region h1 , while the higher excitation line [ oiii]@xmath05007 appears to be brightest in regions a7 and g. these differences in the distribution of the emission line gas are not only due to internal extinction effects , but they also reflect changes in the ionization conditions as a function of location within the galaxy ( see [ ionizinterface ] ) . the stellar light distribution as observed with the hst shows an irregular morphology dominated by the emission coming from the high surface brightness region b2 ( identified therefore as the optical nucleus ) , and from region b1 and complex c ) . ] located at distances of 0.49 and 1.3 kpc from b2 , respectively ( see fig . [ ngcinfo ] ) . however at near and mid - ir wavelengths the true nucleus of the galaxy ( see discussion in [ truenucleus ] ) , b1 , becomes the brightest continuum region in ngc 3690 ( see aah00 ; satyapal et al . this disordered morphology could be the combination of the original spiral structure with young star - forming regions such as the complex c , and regions d3 or c@xmath27 , newly formed as a consequence of the interaction process as suggested by several authors ( e.g. gerhz et al . 1983 ; standford & wood 1989 ; baan & haschick 1990 ; hibbard & yun 1999 ) . the distribution of the ionized gas ( fig . [ ngcinfo ] ) is simpler than that of ic 694 . the optical nucleus ( b2 ) appears as an extremely faint line emitter , and the structure of the ionized gas is dominated by two high surface brightness regions separated by 1.8 kpc , nucleus b1 and the complex c ( with the peak located towards c2 ) , as previously founded with near - ir hydrogen recombination lines ( sugai et al . 1999 ; aah00 ) . the brightest emission line peak is associated to c2 for all lines but the shock tracer [ oi]@xmath06300 . for the latter , the nucleus b1 is the brightest emitter , as also observed with the near - ir [ fe ii]@xmath01.64 @xmath11 m line ( aah00 ) . analyzing in detail the secondary emission line peaks associated to b1 and surrounding areas , it is interesting to note that the high excitation [ o iii]@xmath31 emission is dominated by b1 , while the h@xmath1 and h@xmath25 emission peaks are associated to region b16 ( which is located 0.26 kpc southwest of b1 , and identified in the @xmath04475 continuum map ) . low excitation collisional lines [ n ii]@xmath06584 , [ sii]@xmath216717 , 6731 appear to have secondary peaks in both b1 and b16 . these changes in the emission line distribution represent changes in the ionization conditions , and most likely in the ionizing source ( agn , young stellar clusters and shocks ) on scales of less than 0.3 kpc . as mentioned in the previous section , part of the complex structure observed in the continuum and emission line light distributions is due to large amounts of dust producing spatially dependent internal extinction effects . the two dimensional extinction maps of the gas ( figs . [ ebvinterface ] , [ ebvic ] and [ ebvngc ] ) have been derived using the h@xmath1/h@xmath25 line ratio , a foreground dust screen model , and a mean interstellar extinction law based on that of savage & mathis ( 1979 ) . no correction for the presence of underlying stellar hydrogen absorption lines has been applied . assuming an equivalent width value of ew@xmath32(h@xmath1 ) = ew@xmath32(h@xmath25 ) @xmath6 2 , the derived extinction could be overestimated by as much as @xmath33e(b - v ) @xmath6 0.6 for regions with extremely faint h@xmath25 emission ( e.g. ew(h@xmath25 ) @xmath34 20 ) . the values for the visual extinction ( a@xmath29=3.1@xmath2e(b - v ) ) in particular regions are given in tables [ infoarp ] , [ tabic ] , and [ tabngc ] . the results suggest that the highest extinguished regions are , in general , associated to the nuclei and some particular regions ( like c@xmath27 ) , while the external hii regions are mostly less affected by the presence of dust . the extinction in the whole area covered by the sb3 bundle on scales of about 30 ( i.e. 0.6 kpc ) , is relatively uniform ( see fig . [ ebvinterface ] ) . it has an average value of a@xmath29 = 1.9 mag on the selected individual regions , and ranges from a@xmath29(k2 ) = [email protected] mag to a@xmath29(k5 ) = [email protected] mag ( see table [ infoarp ] for specific values on all the individual regions ) . on scales of about 10 ( i.e. 0.21 kpc ) , the internal extinction in the visual covers a wide range of values from about a@[email protected] mag in the southwestern section of the galaxy to a@[email protected] mag in region h2 , close to the nucleus a ( see table [ tabic ] for specific values , and fig . [ ebvic ] for the two dimensional e(b - v ) distribution ) . as can be seen from fig . [ ebvic ] , the h@xmath1 extinction corrected light distribution shows an overall structure similar to that of the pa@xmath1 line except for the region associated with the peak of the emission . while the nucleus a , identified as the brightest near - infrared source in the hst images ( fig . [ ebvic ] , see also aah00 ) , is also the dominant pa@xmath1 line emitter , the peak of the extinction corrected h@xmath1 distribution is located in regions h1 and h2 , 0.3 and 0.5 kpc towards the east of the nucleus . also region f , located 1.58 kpc to the south - east of a , appears as a strongly absorbed region ( a@[email protected] mag ) , and as a strong h@xmath1 emitter after correcting for extinction . additional secondary h@xmath1 sources are mostly associated with star - forming regions ( the arm , a region towards the west of a ) at distances of 1 to 1.6 kpc from the nucleus , and delineating what seems to be a spiral - like arm structure previously identified in the pa@xmath1 emission line map . this galaxy presents the widest range of extinction ( derived from the h@xmath1/h@xmath25 ratio ) for the entire arp 299 system with values that for the individual regions range from a@xmath29(c4)[email protected] mag to a@xmath29(b2)[email protected] mag ( see table [ tabngc ] ) . also , the nucleus b1 in this galaxy is the region of the arp 299 system which has associated the highest h@xmath1/h@xmath25-based extinction , [email protected] mag ( fig . [ ebvngc ] ) . the overall structure of the extinction corrected h@xmath1 line distribution agrees well with that of the pa@xmath1 line ( fig . [ ebvngc ] , see also aah00 ) not only on the high surface brightness regions such as the nucleus b1 and c1 , but also on the low surface brightness region , including b2 identified as the optical nucleus . additional line ratios involving ir hydrogen recombination lines ( e.g. , pa@xmath1/h@xmath1 , br@xmath28/pa@xmath1 , and pa@xmath1/pa@xmath25 ) , can provide better estimates of the internal extinction in highly extinguished regions as those identified in arp 299 ( see table [ palfatable ] ) . the pa@xmath1 image of the individual galaxies ( see figs . [ ebvic ] , [ ebvngc ] , and aah00 ) has been used for measuring all the regions under study with apertures of 10 for ic 694 and of 09 for ngc 3690 to match our ground - based apertures ; the integral h@xmath1 fluxes are given in tables [ tabic ] and [ tabngc ] . in most regions extinction estimates based on only optical and optical+infrared lines agree within the uncertainties ( table [ palfatable ] ) . on the other hand , there are two regions , the nucleus of ic 694 ( a ) and c@xmath27 , where the use of the optical+infrared lines indicates a larger extinction than using only the optical lines , i.e. a@xmath29(pa@xmath1/h@xmath1)@xmath162@xmath2a@xmath29(h@xmath1/h@xmath25 ) . this means that the extinctions derived using the h@xmath1/h@xmath25 ratio are in general consistent with the ones derived using the pa@xmath1/h@xmath1 ratio , except in highly absorbed regions , where there is an underestimate when only the optical lines are used . these extinctions are similar with those measured in the m51 nuclear hii regions ( calzetti et al . 2005 ) , and in other lirgs ( alonso - herrero et al . 2006 ) using the same near - ir / optical emission lines . the ionization level for the individual regions identified in arp 299 has been derived using standard optical emission line diagnostic diagrams ( bpt diagrams : baldwin , phillips & terlevich 1981 ; veilleux & osterbrock 1987 ) . hii - like ionization ( i.e. star formation ) is present on the majority of the regions of the individual galaxies , except for nucleus b1 which clearly has seyfert - like ( i.e. presence of an agn ) state . liner - like ionization is mainly present on the interface ( fig . [ clasicos ] ; see also tables 1 , 2 , and 3 for specific values of the emission line ratios ) . several regions show changes in the classification depending on the specific diagram used ; these variations have been interpreted as a consequence of changes in the dominant ionizing mechanism from photoionization to shocks ( e.g. monreal - ibero et al 2006 and references therein ) . in particular , the [ oiii]@xmath05007/h@xmath25-[oi]@xmath06300/h@xmath1 line ratio is more sensitive , and therefore a more reliable tracer , to changes in the ionization conditions due to fast shocks than any of the other optical emission line ratios involving [ n ii]@xmath06584/h@xmath1 or [ sii]@xmath06725/h@xmath1 ratios ( dopita & sutherland 1995 ) . in the following subsections a more detailed discussion of the two - dimensional ionization structure of the ionized gas distribution based on the [ oiii]@xmath05007/h@xmath25-[o i]@xmath06300/h@xmath1 diagram and on the [ o iii]@xmath05007/h@xmath25 , and [ oi]@xmath06300/h@xmath1 maps for the interface , ic 694 , and ngc 3690 is given . the [ oiii]@xmath05007/h@xmath25-[oi]@xmath06300/h@xmath1 line ratios have been obtained for all the integral / sb3 individual spectra ( fig . [ diaginterface ] ) . although these data include the high surface brightness nuclear regions of ic 694 and ngc 3690 , and their external regions , they also cover the area that has been identified as the interface zone , i.e. the extended low surface brightness region connecting the two galaxies up to distances of about 4 kpc from their nuclei . changes in the ionization structure traced by the [ oiii]@xmath05007/h@xmath25 and [ oi]@xmath06300/h@xmath1 two dimensional spatial distribution line ratios are also shown ( fig . [ diaginterface ] ) . three different types of ionization , hii- , liner- and seyfert - like , are identified , and the line ratio maps clearly show that the ionization structure has a well defined spatial distribution . the majority of the regions associated with the two individual galaxies ( marked with two iso - contours in the figure ) are dominated by hii - like ionization , as expected from the intense star formation taking place there ( see more details in [ sec9 ] ) . there are some exceptions ; the nucleus b1 shows liner - like activity at this angular resolution . the seyfert - like ionization is clearly resolved with a size of about 7 ( @xmath61.5 kpc ) and has a conical morphology , with an opening angle of about 54@xmath36 and its apex located in nucleus b1 ( at a projected distance of about 1.5 kpc from the peak of the nebula ) within the angular resolution ( see fig.[diaginterface ] ) . moreover , this highly ionized gas is not associated to any particular concentration of stellar mass according to the optical ( wfpc2/f814w ; see fig . [ galaxia ] ) hst continuum image . off - nuclear seyfert - like nebulae at distances of few kpc from the nucleus have already been reported in other ( u)lirgs with seyfert nucleus ( mrk 273 : colina et al . 1999 ) , and interpreted as photoionization of extranuclear interstellar gas by the agn . therefore , our data strongly suggest photoionization by a radiation cone escaping from a central dust - enshrouded agn source located in b1 . the highly ionized conical structure in arp 299 ( ngc 3690 ) is also detected at distances of up to 4 kpc from b1 ; however , the ionization associated to these outer regions of the cone is liner - like rather than seyfert ( fig . [ diaginterface ] ) . therefore , it seems that within the cone there are two ionization regimes well separated spatially . at projected distances from b1 smaller than about 2 kpc the ionization is likely due to radiation coming from the agn located in b1 , while at larger distances liner - like ionization is dominant . this liner - like ionization is also found in other areas of the interface region , however with a lower degree of excitation , as traced by the [ o i]@xmath06300/h@xmath1 ratio measured in these regions . over the area covered by the integral sb2 bundle , hii and liner - like ionization are present in this galaxy ( fig . [ diagic ] ) . the main body of the galaxy , including the regions close to the nucleus a and the spiral arm structure ( a7 , g ) towards the east ( f ) , is dominated by the star - forming knots . liner - like ionization is present elsewhere , being more prominent in diffuse low surface brightness regions located at a projected distance of about 1.8 kpc . the [ nii]@xmath06584/h@xmath1 and the [ sii]@xmath06725/h@xmath1 line ratio maps ( not shown ) also have a similar structure though , as explained above , they tend to indicate a lower ionization . three types of excitation level ( hii- , liner- and seyfert - like ) are present in this individual galaxy , as shown in the [ oiii]@xmath05007/h@xmath25-[oi]@xmath06300/h@xmath1 diagnostic diagram and in the two dimensional maps ( fig . [ diagngc ] ) . regarding the nucleus b1 , these higher resolution data give a clear seyfert - like classification in all these line ratios ( fig . [ clasicos ] ) , and show that there is an extended zone of about 2 ( 0.4 kpc ) in length associated to b1 with seyfert characteristic . this is the inner section of the agn - like ionization cone detected with the sb3 bundle , with its apex coincident with b1 . moreover , the region b2 presents also a high ionization state in the border line of seyfert / liner , characteristic that could indicate the presence of the counter ionization cone . the detection of seyfert - like activity based on optical emission lines represent the first evidence in the optical for the presence of an agn in b1 , and supports the recent claim of an agn based on x - ray ( della ceca et al . 2002 ) and mid - ir ( gallais et al . 2004 ) data . previous spectroscopic studies in the optical ( gehrz et al . 1983 ; coziol et al . 1998 ; keel 1984 ) did not find agn emission , probably due to the large size of the slits ( 36 , 25 , and 80 ) . this is in agreement with our results , where the seyfert nature of the b1 nucleus has been diluted in the sb3 ( = 27 ) spectra , whereas in the higher spatial resolution sb2 ( = 09 ) spectra the classification is clear . hii - like ionization dominates the rest of the main body of the galaxy ( e.g. complex c ) , while liner - like is present in a few outer regions . the [ oi]@xmath06300/h@xmath1 map has its peak displaced about 07 ( 0.14 kpc ) to the north with respect to b1 . similar displacements are observed in the [ nii]@xmath06584/h@xmath1 and in the [ s ii]@xmath06725/h@xmath1 maps . these lines and in particular [ o i]@xmath06300/h@xmath1 are a good tracers of shocks , thus these shifts may indicate different ionization mechanisms around b1 on scales of 200 pc . these small scale ionization changes need further investigation with high angular resolution spectroscopy . the two - dimensional electron density maps of the ionized gas distribution have been derived for the entire arp 299 system using the standard [ s]@xmath06716/[s]@xmath06731 lines ratio ( aller 1984 ; osterbrock 1989 ) . the average line ratio for the interface and the individual galaxies show a similar value , about [email protected] , indicating an average uniform electron density ( @xmath6100 @xmath37 ) over the entire arp 299 system . however , the ionization cone detected in ngc 3690 presents a well defined gradient of this line ratio value , that roughly ranges from [email protected] in the higher excitation ( i.e. seyfert - like ) center of the cone , to [email protected] in the lower excitation ( i.e. liner - like ) outer section . thus , the electron density seems to decrease from about 180 @xmath37 at a distance of 1.5 kpc from b1 to less than 10 @xmath37 at about 4 kpc . most regions of the arp 299 system have been classified as liner - like , although the mechanism producing this ionization needs to be determined . some authors have found in ( u)lirgs a positive correlation between the [ oi]@xmath06300/h@xmath1 ratio and the velocity dispersion ( monreal - ibero et al . 2006 ) , which has been considered as evidence supporting the shocks as the origin for the ionization ( armus et al . 1989 ; dopita & sutherland 1995 ) . the [ oi]@xmath06300/h@xmath1 versus the h@xmath1 velocity dispersion for all the spectra obtained with integral has been represented ( fig . [ correlacion ] ) in order to investigate the origin of the liner - like ionization in arp 299 . the four main regions of the system observed with the integral sb2 and sb3 bundles are analyzed separately : the inner and outer sections of the ionization cone , the interface region and finally the individual galaxies . the outer section of the ionization cone ( stars in fig . [ correlacion ] , _ left panel_. liner - like excitation , i.e.[oi]@xmath06300/h@xmath1 @xmath38 -1.25 ) is independent of the velocity dispersion and present a stable ionization level , while with a wider variety in the ionization level , the inner section of the cone ( seyfert - like , triangles in fig . [ correlacion ] , _ left panel _ ) does not follow the correlation founded in external regions of ulirgs ( monreal - ibero et al . 2006 ) . the interface region ( crosses in fig . [ correlacion ] , _ left panel _ ) present a stable liner - like excitation level , and is not dependent of the velocity dispersion . with regards to the individual galaxies ic 694 and ngc 3690 as observed with the lower angular resolution ( sb3 ) bundle ( filled circles in fig . [ correlacion ] , _ left panel _ ) , the ionization is mainly h - like , with no evidence of a kinematical correlation . for the higher angular resolution ( sb2 ) data ( see fig . [ correlacion ] , _ center and right panels _ ) , the actual measurements of the individual galaxies do not follow the correlation , although they present a well defined clustering . the galaxy ic 694 has a mean log@[email protected] , and presents about the same fraction of h and liner activity . in the case of ngc 3690 the velocity dispersion is clustered around log@xmath23@xmath62 , and there is a large fraction of regions with h - like ionization . clearly the present data show a different behavior than that found in ulirgs ( monreal - ibero et al . 2006 ) , and this is probably due to the fact that on each case the measurements are done over regions located at different distances to the nucleus , and with different characteristics . the original correlation was derived for diffuse extra - nuclear regions in ulirgs , which are located at distances of about 5@xmath2215 kpc from the nucleus of the galaxy . in the present case the interface region and the ionization cone have regions at distances of up to 4 kpc from the nucleus , and comparing with the original limits of the correlation ( see shaded area on fig . [ correlacion ] , _ left panel _ ) the data of the interface and both regions of the ionization cone lie in general above the ratio , i.e. this regions show higher excitation for the same velocity dispersion . in the case of the individual galaxies observed with the actual higher resolution bundle ( fig . [ correlacion ] , _ center and right panels _ ) the present data are closer to the nucleus than in the case of ulirgs ( up to @xmath62.5 kpc ) and in most cases the regions involved are compact bright sources rather than diffuse . comparing with the original limits of the relation ( see shaded area on fig . [ correlacion ] , _ center and right panels _ ) the ic 694 data are clustered around , while the ones of ngc 3690 are below the correlation . this means that in the present case for the individual galaxies the star formation influence is still present , and that the velocity dispersion close to the nucleus traces mass rather than flows / shocks . given an ionization mechanism , the ionization state of the surrounding gas can be changed as a result of the variations in the electron density , that affect to the ionization parameter ( _ _ u__@xmath39n@xmath40@xmath8r@xmath41 ) . this appears to be the case in the outer high excitation cone associated to ngc 3690 . as already mentioned , the electron density seems to have a radial variation within the cone , and by having the exact variation of n@xmath40 with r , the radial dependence of the parameter _ u _ would be described more precisely . unfortunately in the present case the exact radial dependence of n@xmath40 can not be derived ; in spite of that , the variation detected on the electron density , and therefore the liner activity could be accounted for by the influence of an agn ( if n@xmath40 @xmath39 r@xmath42 ) . the variation of n@xmath40 implies a variation on the ionization parameter ( _ u _ @xmath39 r@xmath43 ) , and thus the low excitation is valid if the source is not local . in conclusion , for the arp 299 system studied here there are no clear evidences for tidally induced shocks as found in ulirgs . therefore , the origin for the liner ionization is unclear , and at least in some zones ( i.e. outer parts of the cone ) it could , perhaps , be due to a radiation field similar to that of a low - luminosity agn . it is well known that merging processes between gas rich galaxies can trigger star formation . as predicted by numerical simulations ( mihos & hernquist 1994 ) , the process can vary depending on whether or not the parent disk galaxies have a bulge . the star - forming process at a low level ( sfr@xmath620 m@xmath14 yr@xmath8 ) consequence of the merging of two disk galaxies without a bulge is produced early on in the merging phase , and can be sustained for about 150 myr . on the other hand , in the merger of galaxies with a bulge the starburst ( sfr @xmath6 50@xmath22100 m@xmath14 yr@xmath8 ) is produced later in the merging phase , and lasts for about 50 myr . to investigate the properties of the young stellar populations of a galaxy assuming instantaneous and continuous bursts , synthetic models such as sb99 ( leitherer et al . 1999 ; vazquez & leitherer 2005 ) can be used . in the following the stellar population synthesis model considers a salpeter initial mass function ( imf ) , with lower and upper mass limits of 0.1 m@xmath14 and 100 m@xmath14 , the geneva tracks with high mass loss , and the calzetti extinction law ( calzetti et al . 2000 ) . instantaneous bursts are normalized to 10@xmath3 m@xmath14 , while the continuous star formation is normalized to a star formation rate of 1 m@xmath14 yr@xmath8 . combining the information from the two dimensional distribution of the h@xmath1 equivalent width and luminosity , and the photometric results , parameters such as the extinction to the gas , the age of the burst , and the stellar mass for each individual region can be derived . for the present analysis we use the integral sb2 bundle with a resolution of 09 ( i.e. 0.18 kpc ) , thus in the majority of cases the data include both the contribution from the stellar clusters and from a diffuse component . the aperture photometry as derived by aah00 , using the hst nic1 f110 m , nic2 f160w , nic2 f222 m images is used in the following analysis . in addition optical ( wfpc2/f814w ) photometry over the same regions , using the same apertures , and after performing the background subtraction and the aperture correction ( for the latter see holtzman et al . 1995 ) has been obtained . the apertures used were 09 diameter for the nuclei of ic 694 and ngc 3690 a and b1 , and for the star - forming region b2 , and 05 diameter for the rest of the star - forming regions . the hst magnitudes and colors derived are shown in table [ magtable ] . in addition , the hst magnitudes and colors were modeled by convolving the simulated galaxy spectral energy distribution with the system ( optics , filter and detector ) transmission function . in the following , instantaneous bursts models characterized by the predicted h@xmath1 luminosity and equivalent width , and near - ir colors will be used to study the stellar populations of the h@xmath1 emitting regions under analysis . comparing the measured values with the models , it is possible to derive the age , extinction and mass of the individual regions under study . regarding to the nucleus b1 , and as previously indicated , it is classified as seyfert , and the fraction of the ionizing flux coming from the buried agn in this region is not well determined . therefore the conclusions regarding the star formation properties in b1 based on the h@xmath1 luminosity have to be taken with caution , and represent an upper limit . because of this fact , the nucleus b1 is basically not included in the following discussion , although the properties derived are listed in the tables . the h@xmath1 luminosity in specific regions with large h@xmath1 equivalent width ( fig . [ ew ] ) traces preferentially the presence of young stellar populations ( @xmath445@xmath2210 myr ) . however , the presence of an old underlying population tends to decrease the ew , and therefore to underestimate the age of the young stellar population . in the case of ic 694 and ngc 3690 the respective nuclei a and b1 , and the ngc 3690 brightest optical region b2 present low ew(h@xmath1 ) . the ages derived from the ew(h@xmath1 ) for several individual regions are shown in table [ tablamodic ] . with an age for the nucleus ( a ) of 5.3@xmath210@xmath3 years , ic 694 does not show a wide spread in age for the individual star - forming regions ; almost all are around 5@xmath210@xmath3 years , except for i1 with 5.9@xmath210@xmath3 years and the youngest regions , concentrated towards the east of the spiral - like arm ( f and g , with 3.3 and 3.4@xmath210@xmath3 years respectively ) . the galaxy ngc 3690 shows an average age for the individual regions of @xmath65.5@xmath210@xmath3 years , with the eldest one ( 7.2@xmath210@xmath3 years ) located towards the north of b2 . the youngest sources are located in the northern part of the galaxy , and correspond to complex c. independent estimates ( satyapal et al . 1999 ) also based on instantaneous bursts and characterized by a salpeter imf , derived ages of about 6@xmath210@xmath3 years for the main components of the arp 299 system , that are compatible with the results obtained here . compared to aah00 , who used a different model with gaussian bursts , the present age for b2 is younger , but applying their ew(br@xmath28)@xmath345 to our model , the age for b2 is @xmath457@xmath210@xmath3 years , compatible with the actual calculations . extinction and dust properties of the star - forming regions can be studied using the near - ir colors f110m - f160w ( approximately _ j - h _ ) and f160w - f222 m ( approximately _ h - k _ ) together with the ew(h@xmath1 ) . under the assumption that the extinction to the stars is equivalent to the extinction towards the gas , the ew(h@xmath1 ) value is independent of a@xmath29 ( fig [ genevamodels1 ] ) . we have unreddened both observed colors using a foreground screen model and the extinction derived from the h@xmath1/h@xmath25 ratio , and compared with the models . once corrected for extinction , and taking into account the error bars , in the majority of the regions the values are close to the predicted a@xmath29=2@xmath223 mag track . thus in general the extinction derived from the h@xmath1/h@xmath25 ratio presented in tables [ tabic ] and [ tabngc ] show an offset with respect to the ones predicted by the colors . there are two possible explanations for this discrepancy ; on the one hand the extinction derived from the h@xmath1/h@xmath25 ratio can be underestimated ( see [ sec54 ] ) , on the other hand the discrepancy traces the contribution of hot dust emission . for the highly reddened regions ( e.g. a , b1 ) , the color excess exists and needs to be explained . ] , c1 ) the colors measured ( and corrected by extinction ) are very red . as already mentioned ( and discussed by aah00 ) one of the most plausible explanations for this color excess is the presence of hot dust ( t@xmath38600 k ) , which would mainly affect the f160w - f222 m color . in an attempt to take into account the contribution of the hot dust emission , the @xmath46@xmath47b@xmath48 wavelength modified black - body thermal emission ( hildebrand 1983 ) has been used . the results demonstrate that depending on the temperature ( which has not necessary to be homogeneous over the line of sight ) and the amount of dust , the contribution of hot dust ( @xmath6800 k ) emission could represent , on average , about 80% of the k - band emission ( fig . [ genevamodels1 ] , _ right panel _ ) for the highly extincted regions ( a , b1 , c1 ) , and about 30% for the rest of the regions . this result supports our previous conclusion that the dust is more concentrated towards the nuclei and some particular regions , than in the extra - nuclear star forming regions . an alternative to the presence of a hot dust emission is that the structure of the dust - gas distribution does not follow the simple screen geometry , but it is closer to a clumpy foreground or a mixed dust - star distribution . assuming a model where the gas is homogeneously mixed with the stars , the effect of an a@xmath29=6 mag over the f160w - f222 m and f110m - f160w measured colors will produce bluer colors : about 0.3 and 0.4 mag respectively ( scoville et al . 2000 ; see fig . [ genevamodels1 ] ) . for the extreme case a@xmath29=30 , the corresponding shifts will be 0.6 and 0.8 mag , i.e. in region c1 the f160w - f222 m color will move from the observed 2.67 to 2.07 . thus the colors modeled with this alternative dust geometry would imply that a hot dust contribution is not mandatory in the external less extincted regions , but it is still required to explain the color excess , in the highly extincted regions ( a , b1 and c1 ) . the stellar mass of the young population has been measured using two different indicators , the h@xmath1 luminosity and the m@xmath49 absolute magnitude , and comparing with the predicted values for the ages derived using the ew(h@xmath1 ) . in both cases the h@xmath1/h@xmath25 ratio has been used for correct by extinction ( for a and c@xmath27 the pa@xmath1/h@xmath1 ratio was used too ) . the results obtained using m@xmath49 ( see fig . [ genevamodels2 ] and table [ tablamodic ] ) show that the regions with more young stellar mass content are nucleus a and regions b2 , b21 and c1 ( 460 , 300 , 58 and 81@xmath210@xmath3 m@xmath14 respectively ) . the stellar mass content for the rest of the regions , including the low - surface brightness knots , ranges between 3.1 and 14.0@xmath210@xmath3m@xmath14 ( details on table [ tablamodic ] ) . previous derivations of the stellar mass of the clusters using absolute h - band magnitudes vs h - k ( not corrected for extinction ) with a different stellar populations synthesis model range , from about 1.5@xmath210@xmath50m@xmath14 to 5.0@xmath210@xmath3m@xmath14 ( aah00 ) . using the h@xmath1 luminosity , the young stellar mass content is again concentrated on the nucleus a , and regions b2 , b21 and c1 ( 655 , 679 , 74 and 74@xmath210@xmath3 m@xmath14 respectively ) . for region c@xmath27 , where there is not near - ir photometry , the mass derived using the h@xmath1 luminosity is 115@xmath210@xmath3m@xmath14 . the rest of the derived masses range from 5.7 to 57@xmath210@xmath3 m@xmath14 . comparing the mass in newly formed stars with the derivations of aah00 , there are no significant differences except for b2 where we derive a value two times higher . the masses of the young h@xmath1-emitting regions identified in arp 299 are within the range of masses founded in lirgs ( 3@xmath210@xmath3 - 10@xmath51m@xmath14 , scoville et al . 2000 ; alonso - herrero et al . 2002 ) and are on average about 100 times more massive than the galactic globular clusters ( 10@xmath50m@xmath14 , van den bergh 1995 ) , probably due to the fact that the present regions have conglomerates of less massive clusters . when comparing the stellar masses derived with the h@xmath1 luminosity with the mass obtained from the absolute f160w magnitude , there is a good agreement in the nucleus a , but for the rest of the regions there are differences from a factor @xmath61 to 5 . one of the main reasons for this difference is that in the majority of the cases ( except a and b2 ) an aperture of 05 has been used for the hst cluster photometry , instead the 09 or 10 used for the integral data analysis ( for the ew(h@xmath1 ) and l(h@xmath1 ) estimates ) ; this difference can lead to offsets of a factor 3 in the mass estimates . factoring in the aperture effects , the two methods give the same mean value within a factor 2 . the derived ages of the young h@xmath1-emitting star - forming regions cover a range of between 3@xmath210@xmath3 years to at least 7 - 8@xmath210@xmath3 years old . this is consistent with the methodology used here : since the h@xmath1 traces preferentially the massive young ionizing stellar clusters , population older than 10 myr is difficult to be detected as h@xmath1 decreases . in any case there is further evidence that the star formation in arp 299 has persisted for at least the last 10 - 15@xmath210@xmath3 years ( aah00 ) , with high activity on the star - forming regions . on the basis of this dispersion , we consider that the overall star formation of arp 299 can be modeled by a continuous star formation . under this premise the star formation rate ( sfr ) and the ir luminosity contribution can be derived from the extinction corrected h@xmath1 luminosity , using kennicutt ( 1998 ) expressions , valid for a salpeter imf with mass cutoffs of 0.1 - 100 m@xmath14 : @xmath52 @xmath53 that strictly speaking are only valid for starbursts with ages less than 10@xmath51 years . without taking into account the nucleus b1 ( agn candidate ) , the sfr of the entire system is about 43 m@xmath4 year@xmath8 , and the l@xmath12 is 9.5@xmath210@xmath54 erg s@xmath8 , whereas the luminosity derived from the iras fluxes is 2.2@xmath210@xmath55 erg s@xmath8 , thus the contribution of the studied regions to the ir luminosity is about 45% . there is a lack of about 55% in luminosity , that can be accounted for by the agn located in b1 . previous estimates based on the mid and far - ir spectra of arp 299 assign a 20% of the l@xmath12 coming from b1 and b2 ( charmandaris et al . 2002 ) , while at 11.7 @xmath11 m the percentage of flux coming from b1 and b2 is about 30% ( keto et al . so even considering that the majority of this contribution comes from b1 , there is still a lack of about 25@xmath2235% . this inconsistence is probably due to the fact that the present result is limited by the derived extinction and/or the models . previous studies of arp 299 ( gehrz et al . 1983 ; telesco et al . 1985 ) concluded that this system consists of two spiral galaxies disturbed by the merging process , which are rotating in opposite sense ( augarde & lequeux 1985 ) . more recently hibbard & yun ( 1999 ) have described arp 299 as a structure formed after a prograde - retrograde encounter between ic 694 ( sab - sb ) and ngc 3690 ( sbc - sc ) that took place some 750 myr ago , and having a long tidal tail ( 180 kpc in length ) detected in h. although there are still two structures clearly distinguishable with the two nuclei separated by about 4.6 kpc , the warm gas component has lost its kinematic identity and the velocity field presents a very irregular pattern ( rampazzo et al . 2005 ) . the kinematics of the ionized and the neutral gas is traced by the h@xmath1 emission line and the nai@xmath05890 absorption line respectively ( sparks et al . the ionized gas mean velocities and velocity dispersions were obtained by fitting the h@xmath1+[nii]@xmath216548,6584 complex with a single gaussian velocity component per line ; the same method was applied to the hei@xmath05876+nai@xmath215890 , 5896 complex . the values derived for the individual regions are presented in tables [ kinearp299 ] , [ kineic694 ] and [ kinengc3690 ] . for the velocity measurements , the errors have been calculated taking into account both the systematic uncertainties of the wavelength calibration ( same for all lines ) in the velocity measurement , and the uncertainties in the measurement of the centroid of the line ( that depends on the s / n of the studied line ) . one of the most relevant results obtained is that the two dimensional structure of the ionized and neutral gas velocity field show similar features in both the individual galaxies and the interface region , indicating that the ionized and neutral gas kinematics are coupled . the structure of the h@xmath1 velocity dispersion shows that there is a variation on the average velocity dispersion values : the interface region has the lowest ( 45 km s@xmath8 ) , ic 694 the intermediate ( 65 km s@xmath8 ) and ngc 3690 ( 120 km s@xmath8 ) the maximum values of @xmath23 . the overall velocity field of the ionized gas as observed with the sb3 bundle has a complex and irregular pattern , that does not indicate the presence of virialized simple ( e.g. , rotating ) motions , on scales of 0.6 kpc or larger ( fig . [ kinematinterface ] , _ left _ ) . the structure of this map agrees well with that obtained by rampazzo et al . ( 2005 ) covering a larger area ( fov of 5.9 ) on the sky . the velocities in the northern part are receding , and in the southern part are approaching to the systemic value , defined as the velocity in nucleus a ( the dust - enshrouded nucleus of ic 694 ) with the higher spatial resolution bundle sb2 , v@xmath56=3121@xmath3027 km s@xmath8 ( see section [ seckineic694 ] ) . the corresponding lower spatial resolution sb3 velocity is v@xmath56=3190@xmath3027 km s@xmath8 . the velocity field of the neutral gas obtained from the nad absorption lines , despite of being noisier ( the intensity of the nad can be typically 2% of that of h@xmath1 ) , shares similar structural characteristics with the ionized gas velocity field ( figure [ kinematinterface ] _ center panel _ ) , and presents a systemic velocity of 3057@xmath3034 km s@xmath8 . the ionized gas velocity field presents a peak - to - peak velocity of about 230 km s@xmath8 , while for the neutral gas the variation is smaller , about 170 km s@xmath8 . in both cases the area with the largest approaching velocities is close to b2 . the ionization cone does not show relevant features in the ionized gas velocity field , nor in the neutral gas velocity field . with the exception of the nuclear regions , the ionized gas in the galaxies does not have a distinct kinematics with respect to the interface region ( i.e. , the structure of the velocity field does not correlate with the distribution of mass as traced by the continuum images ) . the velocity gradient of the ionized gas through the interface is about 40 km s@xmath8 kpc@xmath8 , compatible with the h@xmath1 velocity gradients previously measured across ulirgs observed with a slit ( up to 47 km s@xmath8 kpc@xmath8 for galaxies with double nuclei , martin 2006 ) . the velocity dispersion shows distinct features over the entire area ; there are large differences in @xmath23 , that allow us to discriminate spatially between quiet and turbulent gas with macroscopic motions ( fig . [ kinematinterface ] _ right panel _ ) . the so - called interface has an average @xmath23 value of 45 km s@xmath8 that traces no turbulent gas . the low velocity amplitude and velocity dispersion of the interface indicates that the ionized gas could be in a close to quiescent phase or slowly rotating . larger @xmath23 values are measured on the previously defined ionization cone , with the maximum ( 160@xmath3021 km s@xmath8 ) located on the peak of seyfert - like ionization . this spatial coincidence could indicate that in that area the two contributions of the biconic surface of the cone are added , and thus the line measured is wider ( heckman et al . 1990 ) . on the other hand , the maximum velocity dispersion would imply that the largest macroscopic motions , with approaching velocities , are taking place in that highly ionized area . the velocity field of both the ionized and neutral gas presents a similar structure ( fig . [ kinematic694 ] _ left and center panels _ ) , when observed with the sb2 bundle . it has an irregular structure , with the largest receding velocities associated with region i ( 3181@xmath3027 km s@xmath8 for the ionized gas and 3155@xmath3034 km s@xmath8 for the neutral ) , located at 0.42 kpc sw from the nucleus ( a ) . these maps show a general trend with velocities changing from nw ( receding ) to se ( approaching ) , in agreement with the pattern found with the sb3 data in this region . the systemic velocity for the ionized gas is defined as the velocity associated with the nucleus ( region a ) , and it has a value of 3121@xmath3027 km s@xmath8 , consistent with zhao et al . ( 1997 ) estimate based on the h92@xmath1 radio recombination line ( 3120@xmath3050 km s@xmath8 ) . the ionized gas velocity field has a peak - to - peak variation of 290 km s@xmath8 . the neutral gas systemic velocity is 3057@xmath3034 km s@xmath8 , and the peak - to - peak velocity is about 216 km s@xmath8 . although evident structural similarities are found between the ionized and neutral gas ( i.e. same kinematics ) , in general the velocity of the neutral gas presents a blueshift over the entire area with respect to the ionized gas ( about -30@xmath3012 km s@xmath8 over the studied regions ) . this systematic blueshift appears to be real taking into account the uncertainties . in that case , this is a marginal effect , with a lower value than the ones measured in ulirgs using a slit ( martin 2006 ; from about -60 km s@xmath8 to -300 km s@xmath8 ) . in the present analysis the value obtained is derived from ifs data , i.e. has the advantage of the two dimensional distribution versus the use of the slit , where the measurements are limited to a restricted area , probably centered on the nuclear regions . across the galaxy there are gradients of about 110 km s@xmath8 kpc@xmath8 in the ionized gas velocity ( between regions located @xmath62.3 kpc apart i.e. @xmath611 ) , higher than the typical values founded in ulirgs ( up to 60 km s@xmath8 kpc@xmath8 for an individual galaxy , martin 2006 ) when using a slit of 20 long . in closer regions with large velocity variations the gradients can be 150 km s@xmath8 kpc@xmath8 ( between regions i and northern of g , located about 1.4 kpc apart ) . in spite of these gradients , and as discussed below , there are no clear evidence for virialized simple rotating motions . the two - dimensional velocity dispersion map shows a different structure than the velocity field ( fig . [ kinematic694 ] , _ right panel _ ) . for nucleus a , the @xmath23 value is 108@xmath3011 km s@xmath8 , in well agreement with the derivation of shier et al . ( 1996 ) based on the co(2,0 ) near - infrared band at 2.3 @xmath11 m ( 135@xmath3021 km s@xmath8 ) . the maximum value for @xmath23 ( 170@xmath3020 km s@xmath8 ) is located in a region 0.28 kpc towards the north - east of a , where the [ o ] @xmath06300 emission peak is also located . since [ o ] @xmath06300 trace shocks , this high @xmath23 value is a hint of turbulence and shocks . the lowest velocity dispersion ( lower limit 25@xmath303 km s@xmath8 ) corresponds to the area dominated by hii - like ionization , towards the east of the spiral arm ( e.g. a7 , see [ ionizic ] ) . this indicates that the velocity dispersion is likely dominated by the internal kinematics within the stellar clusters , rather than by large regions projected along the line of sight . in this case it is not possible to state that there are clear indication of pure rotation , i.e. the photometric peak does not coincide with the velocity dispersion maximum , and neither of them coincide with the possible kinematic center . the velocity maps show that even if pure rotation does not dominate the galaxy kinematics , rotational movements can be present . like in the previous case , at a resolution of 0.18 kpc the structure of the ionized and neutral gas velocity fields obtained with the sb2 bundle are very similar ( fig . [ kinematngc3690 ] _ left and center panels _ ) . the largest receding velocities are detected towards the north - east of the field ( c@xmath27 , c6 ) , while the largest approaching velocities are located on b2 and towards the south - west . the ionized gas systemic velocity is defined as the one of the nucleus b1 , with a value of 3040@xmath3027 km s@xmath8 . this value coincides with zhao et al . ( 1997 ) measurements using the h92@xmath1 radio recombination line ( 3080@xmath3040 km s@xmath8 ) . for the ionized gas , the peak - to - peak velocity is 257 km s@xmath8 . the systemic velocity measured with the neutral gas is 2976@xmath3032 km s@xmath8 , with a peak - to - peak variation of 217 km s@xmath8 . in both ionized and neutral gas , the velocity variations are similar to the ones measured in ic 694 . as founded in ulirgs ( martin 2006 ) , there is a systemic blueshift of the neutral gas velocity with respect to the ionized gas ( about -24@xmath3013 km s@xmath8 in the studied regions ) . due to its value , lower than the results founded for ulirgs ( from about -60 km s@xmath8 to -300 km s@xmath8 , martin 2006 ) , this blueshift is consider as a marginal effect . across the galaxy there are gradients of about 70 km s@xmath8 kpc@xmath8 in the ionized gas ( between c@xmath27 and d3 ) . this result is close to the gradients derived for ulirgs using a slit ( up to 60 km s@xmath8 kpc@xmath8 for an individual galaxy , martin 2006 ) . the two - dimensional structure of the velocity dispersion ranges between 50@xmath305 and 172@xmath3020 km s@xmath8 ( fig . [ kinematngc3690 ] _ right panel _ ) , and present different characteristics than the two dimensional velocity fields . the nucleus b1 has the maximum velocity dispersion value of the entire system ( 172@xmath3020 km s@xmath8 ) , and regions with high velocity dispersion ( @xmath6160 km s@xmath8 ) are also detected about 1 kpc towards the north and north - west of b1 . looking at the ionized gas velocity field , this secondary peaks of the velocity dispersion are located in regions where there are high velocity variations ( from approaching to receding ) , and thus this @xmath23 would represent a convolution of a large velocity gradient . although the @xmath23-peak coincides spatially with the nucleus , and then it would be tracing the mass ( large @xmath23 implies that this is the most massive galaxy of the system ) , the presence of an agn in b1 could produce a broadening of the line width . like in the case of ic 694 there are no clear evidences for ordered rotating motions ; the photometric and velocity dispersion peaks do not coincide with the possible kinematic center . in conclusion , although the pure rotation does not dominate the kinematics of the galaxy , rotational process would be present . the galaxy ngc 3690 , shows variations on the surface brightness importance of the regions b1 and b2 with wavelength . while in the ultraviolet ( windhorst et al . 2002 ) and in the optical ( this work ) b2 appears brightest than b1 , and therefore has been identified as the nucleus , the reverse is true in the near - ir ( 2.2 @xmath11 m aah00 ) and mid - ir ( gallais et al . besides , the optical kinematical analysis shows that the @xmath23 peak is located in b1 ( see [ kinngc ] ) . this fact along with the high extinction ( a@xmath29=3.7 ) , the hot dust component needed to explain the color excess , and the presence of an agn , lead us to the conclusion that b1 is the true nucleus of the individual galaxy ngc 3690 , as previously discussed by gallais et al . ( 2004 ) with mid - ir spectro - imaging , and concluded by zezas et al . ( 2003 ) using chandra x - ray observations . until now the only mass derivation in this work has been done for the h@xmath1-emitting young star forming regions . the central ionized gas velocity dispersion is a good tracer of the dynamical mass , thus it can be used as an alternative method for mass derivation . assuming virialization , the dynamical mass of the nuclei and brightest regions can be derived as : @xmath57 where @xmath58 is the half mass radius in pc and @xmath23 is the velocity dispersion in km s@xmath8 . assuming a constant value of 1.75 ( see colina et al . 2005 for a detailed discussion ) , and that the effective radius ( r@xmath59 ) of the light distribution traces the half mass radius in each region , it is possible to infer the total dynamical mass . in the present treatment we assume that h@xmath1 is traced by pa@xmath1 , that gives the real size of the emitting regions . in consequence the effective radius of the most important emitting sources is calculated over the hst pa@xmath1 image ; additional estimates using the f110 m image are done for comparison . there are some caveats related to the spatial resolution of the different set of data worth mentioning . for this mass estimate we are dealing with information at different spatial resolutions : on the one hand the velocity dispersion from the integral sb2 bundle data was measured at a 09 ( @xmath6180 pc ) resolution . on the other hand the r@xmath59 over the most relevant arp 299 regions ( i.e. the ones resolved ) was measured using the nicmos1/f110 m image ( 0043/pix , i.e. @xmath69 pc / pix ) and nicmos2/pa@xmath1 image ( 0076/pix , i.e. @xmath616 pc / pix ) . the typical size of the h regions in lirgs is 200 - 300 pc ( alonso - herrero et al . 2002 ) , i.e. several times larger than the effective radius of the stellar clusters and about the size of a sb2 fiber . taking this into account , the clusters will be point sources for integral , and their @xmath23 will be integrated on the fiber . for nucleus a we derive a dynamical mass of about 1.2@xmath210@xmath60m@xmath14 ( table [ tablamasic ] ) which is about 5 times smaller than the derivation done by shier et al . ( 1996 ) using the velocity dispersion measured from the 2.3@xmath11 m co band ; this discrepancy is due to a different estimate of the radius ( r@xmath61/r@xmath62@xmath164 ) . the calculated m@xmath63 implies that the young stellar mass content derived from l(h@xmath1 ) in a is about 40% . the mean value for the dynamical mass of nucleus b1 is 1.5@xmath210@xmath60m@xmath14 ( in good agreement with shier et al . 1996 ) , consequently the upper limit for the young stellar mass content is about 15% . for b2 the mean value of the total mass is 0.6@xmath210@xmath60m@xmath14 ( the same as shier et al . this would imply that about 100% of the mass in b2 is from young stellar population , indicating that the mass derivation done using l(h@xmath1 ) is probably overestimated . for the last two regions ( c1 and c@xmath27 ) , the r@xmath59 gives an upper limit for the dynamical mass ( 0.6 and 0.3@xmath210@xmath60m@xmath14 respectively , the latter according to the @xmath380.1@xmath210@xmath60m@xmath14 from aah00 ) , that implies young stellar mass contents of about 13% and 40% respectively . a detailed ifs study of the lirg arp 299 ( ic 694+ngc 3690 ) system covering a region of [email protected] kpc has been presented . this work illustrates the flexibility and power of the ifs in studying in two dimensions and simultaneously the different physical properties , such as stellar populations , internal dust extinction , ionization structure and gas kinematics , of complex systems like arp 299 . in what follows only the main results of this study are summarized : 1 . the observed overall two dimensional stellar and ionized gas structure differs due to internal extinction effects caused by the dust , and to the different spatial distribution of the diverse ( agn , stars , shocks ) ionizing sources . the stellar continuum and emission line distribution peaks are not spatially coincident but separated up to 1.4 kpc . the peaks of the ionized gas traced by different emission lines ( e.g. hydrogen recombination , high excitation [ oiii ] , and shock - tracer [ oi ] ) are also not coincident indicating that the ionizing mechanisms change on scales of about 200 pc or less . the region b1 is identified as the true nucleus of ngc 3690 . the presence of an agn in b1 is inferred from the detection of a high - excitation seyfert - like conical structure of about 1.5 kpc in size , an opening angle of about 54@xmath36 , and its apex being located in b1 . this region also presents the additional characteristics of a dust -enshrouded nucleus with high extinction ( a@[email protected] ) , red colors ( f160w - [email protected] ) indicating hot dust emission , and high velocity dispersion ( @xmath23=172@xmath3020 km s@xmath8 ) . the observed near - ir colors reveal the presence of hot dust over the entire structure . according to he f160w - f222 m color , the upper limit for the mean hot dust contribution is about 30@xmath5 at 600 k for the majority of the star - forming regions . the nuclei a and b1 , and region c1 are the most affected by the hot dust component , with an upper limit of about 70@xmath5 , 80@xmath5 and 90@xmath5 at 600 k respectively . the high surface brightness regions are mainly dominated by star formation , while the low surface brightness regions are dominated by liner - like ionization process . the h@xmath1-emitting star - forming regions present an extinction of between a@xmath29=1@xmath226 mag , with ages between 3.3 and 7.2@xmath210@xmath3 years , and masses that range between about 6 and 680@xmath210@xmath3 m@xmath14 . the total contribution ( extinction corrected ) of these sources to the bolometric luminosity is about 45% . the structure of the ionized ( h@xmath1 ) and neutral ( nad ) gas two dimensional velocity fields show similar features , indicating that both components are kinematically coupled on scales of hundreds of pc and larger . the velocity fields are very complex , and in general the galaxy does not appear to be dominated by ordered virialized simple ( e.g. rotation ) motions . the morphology of the velocity field does not correlate with the distribution of mass as traced by the continuum images , that is , no spatial coincidence between the kinematic and the photometric centers exists . + on average , the interface region presents the lowest value of velocity dispersion ( @xmath23@xmath645 km s@xmath8 ) , ic 694 presents the intermediate values ( @xmath23@xmath665 km s@xmath8 ) , and ngc 3690 the highest ( @xmath23@xmath6120 km s@xmath8 ) . this indicate that ngc 3690 is the warmest body of the entire system , and probably the most massive . on the contrary , the low velocity amplitude and velocity dispersion of the interface suggests that the ionized gas is slowly rotating or in a close quiescent phase . + the velocity dispersion measurements over the nuclei and the brightest emitting regions , imply a mean value for the dynamical mass of about 1.2@xmath210@xmath64 m@xmath14 and 1.5 @xmath210@xmath64 for nuclei a and b1 respectively , within a mean radius of 58 pc and 28 pc . for regions b2 , c1 and c@xmath27 the mass ranges between @xmath60.2@xmath210@xmath64 m@xmath14 and @xmath60.6@xmath210@xmath64 m@xmath14 , within a radius of about 30 pc . * acknowledgments * + the authors wish to thank dr . hajime sugai , the referee , for his detailed and constructive report . we thank jess maz - apellniz for his help using _ jmaplot_. this work has been supported by the spanish ministerio de educacin y ciencia , under grant bes-2003 - 0852 , projects aya2002 - 01055 & esp2005 - 01480 . 30 aller , l. h. 1984 , physics of thermal gaseous nebulae ( dordretch : reidel ) alonso - herrero , a. , rieke , g. h. , rieke , m. j. , colina , l. , prez - gonzlez , p. g. , & ryder , s. d 2006 , apj , in press , astro - ph/0606186 alonso - herrero , a. , rieke , g. h. , rieke , m. j. , & scoville , n. z. , 2002 , aj , 124 , 166 alonso - herrero , a. , rieke , g. h. , rieke , m. j. , & scoville , n. z. 2000 , apj , 532 , 845 armus , l. , heckman , t. m. , & miley , g. k. 1989 , apj , 347 , 727 arribas , s. et al . 1998 , proc . spie , 3355 , 821 arribas , s. , mediavilla , e. , garca - lorenzo , b. , & del burgo , c. 1997 , apj , 490 , 227 augarde , r. , & lequeux , j. 1985 , a&a , 147 , 273 baan , w. a. , & haschick , a. 1990 , apj , 364 , 65 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , pasp , 93 , 5 ballo , l. , braito , v. , della ceca , r. , maraschi , l. , tavecchio , f. , & dadina , m. 2004 , apj , 600 , 634 bingham , r. g. , gellatly , d. w. , jenkins , c. r. , & worswick , s. p. 1994 spie , 2198 , 56 calzetti , d. et al . 2005 , apj , 633 , 871 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , & storchi - bergmann , t. 2000 , apj , 533 , 682 charmandaris , v. , stacey , g. j. , & gull , g. 2002 , apj , 571 , 282 colina , l. , arribas , s. , & monreal - ibero , a. 2005 , apj , 621 , 725 colina , l. , arribas , s. , & borne , k. d. 1999 , apj , 527 , l13 coziol , r. , torres , c. a. o. , quast , g. r. , contini , t. , & davoust , e. 1998 , apjs , 119 , 239 della ceca , r. et al . 2002 , apj , 581 , 9 dopita , m. a. , & sutherland , r. s. 1995 , apj , 455 , 468 gallais , p. , charmandaris , v. , le floch , e. , mirabel , i. f. , sauvage , m. , vigroux l. , & laurent o. 2004 , a&a , 414 , 845 garca - lorenzo , b. , acosta - pulido , j. a. , & megias - fernndez , e. 2002 , aspc , 282 , 501 gehrz , r. d. , sramek , r. a. , & weedman , d. w. 1983 , 267 , 551 heckman , t. m. , armus , l. & miley , g. k. 1990 , apjs , 74 , 833 hibbard , j. e. , & yun , m. s. 1999 , aj , 118 , 162 hildebrand , r. h. 1983 , qjras , 24 , 267 holtzman , j. a. , et al . 1995 , pasp , 107 , 156 howarth , i. d. , & murray , j. 1988 , dipso a friendly spectrum analysis program ( starlink user note 50 ; chilton : rutherford appleton lab . ) keel , w. c. 1984 , apj , 282 , 75 kennicutt , r. c. 1998 , ara&a , 36 , 189 keto , e. et al . 1997 , apj , 485 , 598 laurent , o. , mirabel , i. f. , charmandaris , v. , gallais , p. , madden , s. c. , sauvage , m. , vigroux , l. , & cesarsky , c. 2000 , a&a , 359 , 887 leitherer , c. , et al . t. m. 1999 , apjs , 123 , 3 mcleod , b. 1997 , in proc . stsci symp . , the hst calibration workshop , ed . s. casertano , r. jederzejewski , c. d. keyes , & m. stevens ( baltimore : stscii ) , 281 mannucci , f. , et al . 2003 , a&a , 401 , 519 martin , c. l. 2006 , astro - ph 0604173 mattila , s. et al . 2005 , iauc , 8473 , 1 mihos , j. c. , & hernquist , l. 1994 , apj , 431 , l9 moshir , m. , et al . 1993 , vizier on - line data catalog : ii/156a monreal - ibero , a. , arribas , s. , & colina , l. 2006 , apj , 637 , 138 neff , s. g. , ulvestad , j. s. , & teng , s. h. 2004 , apj , 611 , 186 origlia l. , & leitherer c. 2000 , aj , 119 , 2018 osterbrock , d. e. 1989 , `` book - review - astrophysics of gaseous nebulae and active galactic nuclei '' ( university science books ) rampazzo , r. , plana , h. , amram , p. , bagarotto , s. , boulesteix , j. , & rosado , m. 2005 , mnras , 356 , 1177 sanders , d. b. , & mirabel , i. f. 1996 , ara&a , 34 , 749 satyapal , s. , watson , d. m. , pipher , j. l. , forrest , w. j. , fischer , j. , greenhouse , m. a. , smith , h. a. , & woodward , c. e. 1999 , apj , 516 , 704 savage , b. d. , & mathis , j. s. 1979 , ara&a , 17 , 73 scoville , n. z. , et al . 2000 , aj , 119 , 991 shier , l. m. , rieke , m. j. , & rieke , g. h. 1996 , apj , 470 , 222 sparks , w. b. , carollo , c. m. , & macchetto , f. 1997 , apj , 486 , 253 stanford , s. a. , & wood , d. o. s. 1989 , apj , 346 , 712 sugai , h. , davies , r. i. , malkan , m. a. , mclean , i. s. , usuda , t. , & ward , m. j. 1999 , apj , 527 , 778 telesco , c. m. , decher , r. , & gatley , i. 1985 , apj , 299 , 896 van den bergh , s. 1995 , nature , 74 , 215 vazquez , g. a. & leitherer , c. 2005 , apj , 621 , 695 veilleux , s. , & osterbrock , d. e. 1987 , apjs , 63 , 295 windhorst , r. a. et al . 2002 , apjs , 143 , 113 zezas , a. , ward , m. j. , & murray , s. s. 2003 , apj , 594 , l31 zhao , j. , anantharamaiah , k. r. , goss , w. m. , & viallefond , f. 1997 , apj , 482 , 186 lccccccccc k1 & 2.97 & -17.6 & -12.7 & 1.7 & 2.2 & 0.27 & -1.07 & -0.34 & -0.37 + k2 & 3.81 & -21.6 & -13.7 & 1.1 & 0.9 & 0.07 & -1.04 & -0.36 & -0.38 + k3 & 1.28 & -9.6 & -10.7 & 0.6 & 2.0 & 0.22 & -0.59 & -0.21 & -0.25 + k4 & 2.49 & -13.0 & -1.8 & 2.7 & 1.6 & 0.04 & -0.83 & -0.41 & -0.31 + k5 & 3.53 & -14.5 & 3.9 & 10.7 & 2.9 & 0.14 & -1.44 & -0.54 & -0.64 + k6 & 3.71 & -13.1 & 6.0 & 5.7 & 2.3 & -0.05 & -1.27 & -0.48 & -0.54 + k7 & 2.72 & -10.6 & 1.9 & 4.0 & 2.0 & 0.01 & -1.10 & -0.47 & -0.45 + k8 & 4.07 & -9.0 & 9.8 & 1.4 & 2.7 & 0.14 & -0.71 & -0.32 & -0.23 + k9 & 3.52 & -2.6 & 7.8 & 3.3 & 1.8 & -0.01 & -0.92 & -0.37 & -0.36 + k10 & 5.26 & -21.5 & -9.0 & 4.2 & 1.0 & 0.17 & -1.36 & -0.53 & -0.56 + k11 & 1.44 & -9.1 & -5.1 & 1.6 & 1.5 & 0.02 & -0.82 & -0.39 & -0.31 + k12 & 3.18 & -16.8 & -17.4 & 7.6 & 2.0 & 0.13 & -0.98 & -0.35 & -0.37 [ infoarp ] lccccccccc a & 0.00 & 0.0 & 0.0 & 2.9 & 2.9 & -0.14 & -1.03 & -0.31 & -0.41 + a1 & 0.46 & 1.5 & -1.6 & 2.6 & 2.3 & -0.22 & -1.29 & -0.40 & -0.50 + a2 & 0.54 & -1.0 & -2.4 & 3.1 & 2.0 & -0.14 & -1.51 & -0.52 & -0.60 + a3 & 1.37 & 3.3 & -5.7 & 2.3 & 1.9 & -0.03 & -1.61 & -0.53 & -0.59 + a4 & 1.51 & 3.8 & -6.3 & 3.0 & 2.1 & -0.02 & -1.73 & -0.55 & -0.66 + a5 & 1.06 & -2.1 & -4.7 & 6.0 & 1.9 & 0.10 & -1.56 & -0.54 & -0.62 + a6 & 1.00 & -3.6 & -3.9 & 4.0 & 2.1 & 0.04 & -1.46 & -0.52 & -0.56 + a7 & 0.99 & -2.8 & -3.9 & 6.2 & 2.3 & 0.09 & -1.61 & -0.56 & -0.66 + f & 1.58 & 6.5 & -4.1 & 6.7 & 3.2 & 0.10 & -2.06 & -0.52 & -0.92 + g & 1.10 & 1.1 & -5.2 & 6.2 & 1.9 & 0.05 & -1.84 & -0.54 & -0.74 + h1 & 0.48 & 2.3 & -0.3 & 6.4 & 2.8 & -0.15 & -1.41 & -0.47 & -0.63 + h2 & 0.34 & 1.6 & 0.5 & 4.7 & 3.4 & -0.06 & -0.99 & -0.32 & -0.46 + i & 0.42 & -1.8 & -0.9 & 6.0 & 1.6 & -0.09 & -1.60 & -0.52 & -0.65 + i1 & 1.45 & -7.0 & 0.8 & 0.8 & 2.4 & -0.10 & -1.28 & -0.43 & -0.42 + [ tabic ] lccccccccc b1 & 0.00 & 0.0 & 0.0 & 9.7 & 3.7 & 0.55 & -1.07&-0.29&-0.54 + b2 & 0.49 & -1.5 & 1.9 & 4.0 & 3.9 & 0.43 & -1.37&-0.35&-0.74 + b11 & 0.22 & 0.27 & -1.0 & 5.7 & 2.0 & 0.10 & -1.37&-0.41&-0.52 + b16 & 0.26 & -0.5 & -1.1 & 10.6 & 1.1 & -0.02 & -1.68&-0.47&-0.71 + b21 & 0.82 & -1.0 & 3.9 & 3.1 & 2.6 & 0.02 & -1.54&-0.48&-0.77 + b22 & 0.63 & -2.9 & 1.0 & 3.2 & 2.1 & 0.02 & -1.51&-0.41&-0.68 + b25 & 0.85 & -3.9 & 1.3 & 7.2 & 1.7 & -0.02&-1.88&-0.47&-0.80 + c1 & 1.83 & -2.4 & 8.6 & 28.1 & 3.0 & 0.13 & -1.87&-0.49&-0.96 + c2 & 1.80 & -3.0 & 8.2 & 33.4 & 2.1 & 0.11 & -1.92&-0.46&-0.94 + c3 & 1.75 & -3.9 & 7.6 & 16.7 & 1.0 & 0.01 & -1.80&-0.44&-0.78 + c4 & 1.39 & -4.2 & 5.3 & 5.7 & 0.5 & -0.03&-1.80&-0.47&-0.67 + c6 & 1.55 & 1.0 & 7.5 & 5.2 & 1.9 & 0.02 & -1.74&-0.53&-0.70 + c@xmath65 & 1.96 & 2.7 & 9.5 & 0.9 & 3.8 & 0.06 & -1.40&-0.45&-0.57 + d3 & 1.49 & -7.2 & -0.9 & 1.3 & 1.3 & -0.03&-1.60&-0.44&-0.55 + d4 & 1.13 & -5.5 & 0.1 & 2.6 & 2.0 & -0.05&-1.44&-0.41&-0.57 [ tabngc ] lccc a & 167 & 5.8 & 6.1 + a1 & 2.74 & 0.1 & @xmath22 + a2 & 5.68 & 0.2 & 0.9 + a3 & 4.37 & 0.2 & 0.9 + a4 & 9.62 & 0.3 & 1.5 + a5 & 8.41 & 0.1 & @xmath22 + a6 & 2.73 & 0.1 & @xmath22 + a7 & 10.5 & 0.2 & 0.9 + h1 & 12.6 & 0.2 & 0.9 + h2 & 17.3 & 0.4 & 2.0 + i & 5.11 & 0.1 & @xmath22 + i1 & @xmath22 & @xmath22 & @xmath22 + b1 & 66.7 & 0.7 & 2.8 + b2 & 6.10 & 0.2 & 0.9 + b11 & 9.77 & 0.2 & 0.9 + b16 & 18.5 & 0.2 & 0.9 + b21 & 4.14 & 0.1 & @xmath22 + b22 & 5.15 & 0.2 & 0.9 + b25 & 11.0 & 0.2 & 0.9 + c1 & 85.3 & 0.3 & 1.5 + c2 & 76.3 & 0.2 & 0.9 + c3 & 10.0 & 0.1 & @xmath22 + c4 & 4.70 & 0.1 & @xmath22 + c6 & 9.16 & 0.2 & 0.9 + c@xmath27 & 38.1 & 4.1 & 5.6 + d3 & 0.15 & 0.01 & @xmath22 + d4 & 4.81 & 0.2 & 0.9 [ palfatable ] lccccc a & 16.85 & 13.72 & 1.33 & 1.36 & 1.80 + a1 & 18.53 & 17.59 & 1.07 & 1.04 & -0.13 + a2 & 18.57 & 17.36 & 0.91 & 0.57 & 0.30 + a3 & 18.40 & 17.75 & 0.81 & 0.65 & -0.15 + a4 & 18.84 & 18.09 & 1.05 & 0.69 & -0.31 + a5 & 18.19 & 17.61 & 0.72 & 0.63 & -0.14 + a6 & 17.87 & 17.85 & 0.70 & 0.54 & -0.68 + b1 & 15.67 & 13.93 & 1.23 & 2.06 & 0.51 + b11 & 17.30 & 16.31 & 0.88 & 0.64 & 0.11 + b16 & 16.90 & 18.12 & 0.19 & @xmath22 & -1.42 + b2 & 14.09 & 13.19 & 0.70 & 0.69 & 0.20 + b21 & 16.16 & 15.35 & 0.79 & 0.55 & 0.03 + b22 & 17.78 & 16.49 & 0.93 & 0.75 & 0.36 + b25 & 18.01 & 17.67 & 0.94 & 0.62 & -0.56 + c1 & 17.04 & 14.68 & 1.71 & 2.67 & 0.66 + c2 & 16.63 & 16.57 & 0.67 & @xmath22 & -0.61 + c3 & 17.04 & 16.37 & 0.70 & 0.52 & -0.02 + c4 & 17.24 & 16.71 & 0.82 & 0.44 & -0.29 + d3 & 18.74 & 18.37 & 0.70 & 0.46 & -0.33 [ magtable ] lccccc a & 9.2 ( 150 ) & 201 & 5.3 & 40.0 ( 655.0 ) & 200.0 ( 460.0 ) + a1 & 4.8 & 217 & 5.2 & 17.6 & 6.0 + a2 & 4.6 & 292 & 5.0 & 13.0 & 6.4 + a3 & 3.2 & 297 & 5.0 & 9.0 & 4.8 + a4 & 4.7 & 382 & 4.8 & 10.0 & 4.0 + a5 & 7.3 & 307 & 5.0 & 21.0 & 5.3 + a6 & 6.1 & 197 & 5.3 & 26.8 & 5.0 + a7 & 11.3 & 277 & 5.1 & 36.5 & @xmath22 + f & 28.6 & 760 & 3.3 & 21.8 & @xmath22 + g & 7.9 & 654 & 3.4 & 6.2 & @xmath22 + h1 & 18.6 & 416 & 4.7 & 34.0 & @xmath22 + h2 & 22.5 & 314 & 4.9 & 57.0 & @xmath22 + i & 5.8 & 220 & 5.2 & 21.5 & @xmath22 + i1 & 1.6 & 86 & 5.9 & 12.9 & @xmath22 + b1 & 62.3@xmath66 & 240@xmath66 & 5.2@xmath66 & 231.0@xmath66 & 200.0@xmath66 + b2 & 29.8 & 22 & 7.2 & 678.0 & 300.0 + b11 & 7.9 & 200 & 5.3 & 34.6 & 14.0 + b16 & 7.1 & 311 & 5.0 & 20.0 & 3.1 + b21 & 7.6 & 62 & 6.1 & 74.0 & 58.0 + b22 & 4.8 & 138 & 5.6 & 27.3 & 10.0 + b25 & 7.7 & 449 & 4.6 & 13.7 & 5.6 + c1 & 93.8 & 680 & 3.4 & 74.0 & 81.0 + c2 & 53.4 & 645 & 3.5 & 43.8 & 9.8 + c3 & 9.7 & 440 & 4.7 & 17.7 & 9.8 + c4 & 2.1 & 156 & 5.4 & 10.3 & 8.4 + c6 & 6.8 & 609 & 3.6 & 5.7 & @xmath22 + c@xmath65 & 5.2 ( 31.0 ) & 245 & 5.2 & 19.0 ( 115.0 ) & @xmath22 + d3 & 1.0 & 116 & 5.7 & 6.4 & 3.2 + d4 & 3.6 & 360 & 4.8 & 7.8 & @xmath22 [ tablamodic ] lccc k1 & -56 & -29 : & 61 + k2 & -32 & @xmath22 & 58 + k3 & -65 & 30 & 80 + k4 & 13 & 31 & 68 + k5 & 54 & 101 & 35 + k6 & 54 & 115 & 20 + k7 & 36 & 78 & 30 + k8 & 84 & 42 & 70 + k9 & 20 & -6 : & 65 + k10 & 54 & 86 : & 100 + k11 & -2 & 26 & 19 + k12 & -78 & @xmath22 & 36 [ kinearp299 ] lccc a & 0 & 0 & 108 + a1 & -86 & -47 & 73 + a2 & -7 & 22 & 64 + a3 & -134 & -86 & 36 + a4 & -151 & -96 & 50 + a5 & -67 & -25 & 45 + a6 & -44 & -37 & 40 + a7 & -57 & -26 & 41 + f & -163 & @xmath22 & 72 + g & -121 & -77 & 43 + h1 & -130 & -89 & 61 + h2 & -117 & -93 & 109 + i & 60 & 98 & 61 + i1 & 37 & @xmath22 & 79 [ kineic694 ] lccc b1 & 0 & 0 & 172 + b2 & -116 & -55 & 110 + b11 & 38 & 50 & 143 + b16 & -46 & -15 & 117 + b21 & -18 & 9 & 128 + b22 & -121 & -48 & 78 + b25 & -112 & -64 & 54 + c1 & 32 & 65 & 93 + c2 & 16 & 51 & 99 + c3 & 13 & 36 & 90 + c4 & -24 & 8 & 94 + c6 & 115 & @xmath22 & 82 + c@xmath65 & 101 & @xmath22 & 69 + d3 & -101 & -40 & 67 + d4 & -112 & -52 & 54 [ kinengc3690 ] lcc a & [email protected] & [email protected] + b1 & [email protected] & [email protected] + b2 & [email protected] & [email protected] + c1 & [email protected] & [email protected] + c@xmath65 & [email protected] & [email protected] + [ tablamasic ]	the luminous infrared galaxy arp 299 ( ic 694+ngc 3690 ) is studied using optical integral field spectroscopy obtained with the integral system , together with archival _ hubble space telescope _ wfpc2 and nicmos images . the stellar and ionized gas morphology shows @xmath0-dependent variations due to the combined effects of the dust internal extinction , and the nature and spatial distribution of the different ionizing sources . the two - dimensional ionization maps have revealed an off - nuclear conical structure of about 4 kpc in length , characterized by high excitation conditions and a radial gradient in the gas electron density . the apex of this structure coincides with b1 region of ngc 3690 which , in turn , presents seyfert - like ionization , high extinction and a high velocity dispersion . these results strongly support the hypothesis that b1 is the true nucleus of ngc 3690 , where an agn is located . in the circumnuclear regions hii - like ionization dominates , while liner - like ionization is found elsewhere . the h@xmath1 emitting sources with ages from 3.3 to 7.2@xmath210@xmath3 years , have masses of between 6 and 680@xmath210@xmath3 m@xmath4 and contribute ( extinction corrected ) about 45@xmath5 to the bolometric luminosity . the ionized ( h@xmath1 ) and neutral ( nad ) gas velocity fields show similar structure on scales of several hundred to about 1 kpc , indicating that these gas components are kinematically coupled . the kinematic structure is complex and on scales of about 0.2 kpc does not appear to be dominated by the presence of ordered , rotational motions . the large velocity dispersion measured in ngc 3690 indicates that this galaxy is the most massive of the system . the low velocity amplitude and dispersion of the interface suggest that the ionized gas is slowly rotating or in a close to quiescent phase .
phenomenon of electrochromism is found in many materials such as : organic compounds @xcite , polymer composites @xcite and semiconductors @xcite . one of the most studied materials of this class is tungsten trioxide due to a fortunate combination of its corrosion resistivity , sufficient light absorption and satisfactory mechanical properties that make it an attractive material for advanced applications @xcite such as `` smart windows '' technology @xcite . the first mentioning of @xmath5 coloration upon ion insertion dates back to the 19th century @xcite . in the 20th century the interest to this material and its electrochromic properties revived and resulted in a large number of studies @xcite . it is well known by now that `` prussian blue '' coloration in the crystalline and amorphous phases of tungsten trioxide appears upon hydrogen or alkali metal ( @xmath6 ) intercalation @xcite . tungsten trioxide is a wide band - gap semiconductor ( 2.6 - 3.0 ev @xcite)with perovskite like structure , see fig.1 . the oxide shows rich structural polymorphism @xcite , whereby oxygen octahedrons surrounding tungsten atoms vary their orientations and shape depending both on temperature and pressure . usually , at room temperature both @xmath7-@xmath5 ( triclinic @xmath8 233 - 290 k ) and @xmath9-@xmath5 ( monoclinic @xmath10 290 - 350 k ) phases coexist@xcite . depending on the method of preparation samples can be characterized by various @xmath9/@xmath7 ratios , micro - structure and defectiveness . under certain conditions a more exotic hexagonal h-@xmath11 phase @xcite or amorphous a-@xmath11 @xcite phase can form . the structural variety of @xmath5 offers a range of optical characteristics that is important for applications relying on light absorption @xcite . alkali metal intercalation can also cause structural transformation of @xmath5 . for example , in the case of @xmath12 several phase transitions from monoclinic to cubic symmetry with increasing @xmath13 content have been reported @xcite . it has also been shown that for @xmath14 intercalation the degree of coloration increases proportionally to the lithium concentration whereas in the case of hydrogen insertion a saturation level of coloration is reached for a certain @xmath15 concentration @xcite . based on neutron diffraction data @xcite this difference between @xmath16 and @xmath13 has been explained by the formation of @xmath17 bonds upon hydrogen insertion . at the same time alkali metal ions show no tendency of binding to the oxygen atoms of the oxide matrix . substoichiometric @xmath18 also shows multifarious electrochromic behavior , demonstrating different degrees of coloration depending on the concentration of oxygen vacancies @xcite . thus , experiments clearly indicate that both doping ( @xmath19 ) and oxygen vacancies can influence @xmath20 optical properties @xcite . two general concepts of the colouration mechanism in tungsten oxide have been suggested : inter - valence charge transfer @xcite and polaron models @xcite , the latter being more commonly accepted . according to the polaron model the coloration is the result of photon absorption due to photon activated polaron hopping . a polaron can form in @xmath5 due to electron localization at the @xmath21-states of tungsten , thus , turning @xmath22 into @xmath4 or @xmath23 . light absorption can lead to the activation of the following transitions : @xmath24 or @xmath25 . we notice that more complex configurations of localized charge can form , for example , two bound polarons at neighboring sites , @xmath4-@xmath4 , also called bipolarons ( see fig.1b , c for illustration ) . the presence of different localized charge states in @xmath5 has been confirmed by numerous experiments @xcite . the formation of polarons in this oxide has also been studied theoretically by semi - empirical models @xcite and more recently by ab - initio methods employing hybrid functionals @xcite . more studies of the electronic and structural properties of perfect and defective @xmath26 can be found in refs . despite intensive research many questions regarding the properties of polarons in @xmath26 remain unclear . in this paper we address some of them , in particular , we study the formation and mobility of polarons in two cases : 1 ) in the presence of an oxygen vacancy and 2 ) in the presence of li . we investigate the relative stability of polarons and bipolarons and calculate the energy barriers for polaron propagation in @xmath20 . [ cols="<,<,<,<",options="header " , ] [ tab : tb1 ] the monoclinic phase of @xmath5 ( @xmath27 ) has two tungsten and six oxygen non - equivalent atoms @xcite , as illustrated in fig.1a . the structure is rather complex , each @xmath28 atom has two neighboring @xmath29 atoms but the @xmath30 chains differ in all the directions . along [ 100 ] we find @xmath31 and @xmath32 chains , along [ 001 ] @xmath33 and @xmath34 chains interconnected in the [ 010 ] direction by alternating @xmath35 and @xmath36 chains ( for notations see fig.1a ) . the @xmath37 bonds in all the directions have alternating shorter and longer distances and slightly different @xmath38 angles , which are shown in table ii , both calculated with @xmath39 and from the experimental data @xcite . vacancy formation energies for different vacancy positions and two electronic configurations : @xmath40 and @xmath41 . the energies are shown with respect to the lowest formation energy , the one for the @xmath42 position , with the energy of 2.36 ev . see also table iii . ] energies along the transition path between the @xmath41 and @xmath40 configurations for three vacancies ( @xmath43 , @xmath44 and @xmath42 ) as estimated by @xmath45 . the energies are given with respect to that of @xmath41 in the corresponding direction . for comparison for @xmath42 we show the results obtained by the linear interpolation method @xmath46 @xcite . charge distributions corresponding to the initial x=0 ( @xmath41 ) and final x=1 ( @xmath23 ) transition points are shown for illustration . ] using the structure obtained in @xmath39 we calculated the vacancy formation energetics for the six oxygen positions ( see fig.1a ) and two bipolaronic configurations : @xmath40 and @xmath41 bipolaron ( see fig.1b ) . both electronic configurations could be stabilized for all the vacancies . the results are presented in fig.3 and table iii . the lowest vacancy formation energy , 2.36 ev , is found for the oxygen vacancy at site @xmath42 with bipolaronic configuration @xmath41 and the highest energy , 2.90 ev , for the @xmath47 position in the @xmath40 configuration . the formation energies for the @xmath40 configurations are systematically higher than those for the @xmath41 ones ( table iii ) . the vacancy formation energy , @xmath48 , depends on the orientation of the bipolaron in the lattice , @xmath48 averaged for each direction can be sorted in the following order : @xmath49>e_{v}[010]>e_{v}[001]$ ] . these results are in agreement with previous studies carried out by standard @xmath50 @xcite , where the oxygen atoms with the @xmath37 bonds along the @xmath3 $ ] direction were indicated as favorable sites for vacancy formation . using the @xmath51 hybrid functional wang et al . @xcite also reported the formation of the @xmath41 configuration aligned with @xmath3 $ ] . the data in table iii show that the vacancy formation energy decreases as the @xmath52 distances increase and lowest @xmath53 corresponds to the largest @xmath52 separation . the formation of bipolarons ( @xmath41 ) leads to a noticeable increase in the @xmath52 bond lengths due to coulomb repulsion . these local lattice distortions are , however , different for different crystallographic directions . in the [ 100 ] and [ 010 ] directions the @xmath52 distance increase is of about 0.20 whereas for the @xmath54 and @xmath42 vacancy positions ( @xmath3 $ ] ) it is 0.42 and 0.44 , respectively . we see a similar trend for the @xmath40 polaronic configurations ( table iii ) , however , the changes of the @xmath52 distance are smaller in this case . these results indicate certain softening of phonon modes along @xmath3 $ ] . they also indicate that in the case of monoclinic @xmath5 the electronic contribution to the energy of vacancy formation dominates over the elastic contribution . our calculations suggest that the bipolaronic states associated with an oxygen vacancy are tightly trapped as all our attempts to localize charge away from the vacancy were unsuccessful . therefore , we can expect that in the presence of oxygen vacancies bipolarons are immobilized in the lattice , however , the transitions between vacancy bound @xmath41 and @xmath40 could still contribute to the coloration effect in @xmath5 . to study the transition between these two configurations we chose oxygen vacancies with lowest @xmath53 in each crystallographic direction , namely , @xmath43 , @xmath44 and @xmath42 . our results , presented in fig.4 , show that the transition barriers vary from 150 to 230 mev , being the smallest for the @xmath55 $ ] direction and largest for @xmath3 $ ] . this result is in agreement with similar trends shown by the energy differences between the @xmath41 and @xmath40 configurations : @xmath56 and the @xmath52 bond length differences : @xmath57 . we also notice that the @xmath45 relaxation results in lower barriers compared to those obtained with @xmath58 ( fig.4 ) . this is understandable as @xmath45 allows one to optimize structures along the transition path whereas @xmath59 uses static configurations . lattice distortions around polarons and especially bipolarons can be quite substantial reaching far beyond first coordination shells . in the case studied here we find that the local distortions around the @xmath60 complex spreads as far as 10 - 11 in the direction of the @xmath41 pair alignment . the 2x2x2 cell used in our calculations is big enough to accommodate such a defect . our results , however , highlight the importance of using large unit cells in modeling polarons and bipolarons at least if one wishes to study isolated quasiparticles of the sort . polarons can also form in the presence of impurities , dopants or due to electron injection . here we report our results of the polaron formation and mobility in @xmath13 doped monoclinic @xmath5 . first we consider one lithium atom situated in an octapore in the supercell of 256 atoms , i.e. @xmath62 . this is a realistic concentration often studied in experiment @xcite . after the relaxation li stayed close to the octapore center ( fig 1c ) , no bond formation with the surrounding oxygen atoms was detected . it was possible to localize @xmath4 polaron at different tungsten sites . the association energy of @xmath14 and @xmath4 is of the order of 30 mev , which was estimated by calculating the energies of @xmath63 separated by 3.92 and 7.71 . further , to study polaron mobility we used tungsten sites away from @xmath13 ( @xmath64 6 - 8 ) simulating a free polaron propagation in the @xmath5 matrix . fig.5 shows the energy barriers for polaron jumping between two neighboring w sites in the three crystallographic directions . as expected the @xmath59 barriers are again higher than those obtained using @xmath45 . the @xmath45 barriers vary from 98 for @xmath3 $ ] to 124 mev for @xmath55 $ ] . these values are in surprisingly close agreement with the activation energy for polaron hopping estimated from basic models using experimental information , 180 mev @xcite . if the concentration of lithium ions and polarons increases the number of possible configurations of localized charges becomes larger . we study the possibility of a bipolaron formation by considering two li atoms in the 256 atom @xmath5 supercell ( @xmath65 ) . next to the bipolaronic @xmath41 configurations we also consider the formation of two separate @xmath4 polarons and one @xmath23 . in fig.6 we show the energies for different electronic configurations and transitions between them obtained using @xmath45 . we start from the situation when both electrons donated by the two @xmath13 atoms are delocalized in the oxide matrix ( x=0 , fig.6 ) , then we localize one @xmath4 polaron that happens without any barrier ( x=2 , fig.6 ) . after that we localize the second electron at the next nearest @xmath66 @xmath29 site from first @xmath4 , which neither involves any barrier(x=3 , fig.6 ) . the energies of such @xmath67 configurations are very close to each other for all the orientations ( fig.6 ) . next we model a transition of one @xmath4 from the @xmath67 site to the nearest @xmath68 @xmath29 site , thus , forming a bipolaron along the @xmath55 $ ] , @xmath69 $ ] and @xmath3 $ ] directions ( x=3 @xmath70 x=5 , fig.6 ) . this transition involves a barrier of about 90 mev . the resulting bipolaron situated along @xmath3 $ ] has the lowest energy not only among the bipolarons but among all the considered configurations ( fig.6 ) . the @xmath3 $ ] bipolaron is 8 mev lower in energy than the @xmath3 $ ] @xmath67 configuration . at the same time , the energies of bipolarons oriented along @xmath55 $ ] and @xmath69 $ ] are about 20 mev higher than those of the @xmath67 configurations . we notice that the localization of two electrons at one tungsten site ( @xmath23 ) is energetically unfavorable as this state lies more than 300 mev above any @xmath41 configuration . therefore , our results suggest that in @xmath13 doped @xmath5 one can expect the formation of @xmath4 polarons and @xmath41 bipolarons whereas the presence of the @xmath23 state , often detected in experiments @xcite , most likely can be attributed to oxygen vacancies , around which the @xmath23 state is found to be metastable . energies along the transition paths of @xmath4 from one tungsten site to the other in the case of @xmath13 doped @xmath61 phase . the energies are given with respect to the energy of @xmath4 . the solid lines show the @xmath45 energies for all the three crystallographic directions . for comparison , for @xmath55 $ ] we also show the energies obtained in @xmath59 calculations ( dashed line ) . charge distributions for @xmath4 at the initial @xmath71 and final @xmath72 transition points are shown for illustration . ] we have studied the applicability of @xmath73 and two @xmath74 approaches to modelling of polarons in @xmath5 . we have shown that the @xmath74 approach with two @xmath75 parameters determined and applied simultaneously to the d - w ( @xmath76=6 ev ) and p - o ( @xmath77=9 ev ) states provides a proper description of polaron formation and transitions in @xmath5 . at the same time , this method allows us to calculate large unit cells necessary for adequate modelling of polarons and bipolarons . using this approach we have studied the vacancy energetics for the six non - equivalent oxygen positions and two electronic configurations : @xmath41 and @xmath40 . the @xmath41 bipolarons situated along @xmath3 $ ] are the most favorable electronic configurations around vacancies . we have also studied polaron and bipolaron formation and energetics for li doped @xmath5 . in this case the @xmath41 bipolaronic configuration aligned with @xmath3 $ ] has again the lowest energy but winning only 8 mev over two separated @xmath4 . our results suggest that the polarons formed due to oxygen vacancies are immobile , at the same time , the @xmath23 state is metastable and the @xmath41 to @xmath40 transition is possible with a barrier of 150 mev . on the contrary , polarons formed in @xmath13 doped tungsten oxide are mobile with the minimum activation energy ( 98 mev ) in the [ 001 ] direction . the @xmath23 state is 300 mev higher in energy than any studied @xmath4 configuration and , therefore , @xmath23 is unlikely to form in perfect @xmath78 without presence vacancies or similar structural defects . we acknowledge the financial support by the swedish energy agency ( energimyndigheten , stem ) essence , standupp and the swedish research council ( vetenskapsrdet ) . supercomputer time was granted by the swedish national infrastructure for computing ( snic ) . n.b . is grateful to national supercomputer centre ( nsc ) for provided support . o.e . acknowledges in addition the kaw . authors are thankful to c - g . granqvist and g. niklasson for valuable discussions . mortimer , electrochim . acta * 44 * , 2971 ( 1999 ) f. carpi , d. de rossi , ieee t inf . technol . b * 9 * , 295 ( 2005 ) j. heinze , electronically conducting polymers , top . chem . * 152 * , ( 1990 ) c.g . granqvist , handbook of inorganic electrochromic materials , elsevier , ( 1995 ) g. hodes , d. cahen , j. manassen , nature * 260 * , 312 ( 1976 ) a. azens et al . , solid state ionics * 86 * , 943 ( 1996 ) h. watanabe et al . , chem . commun . * 49 * , 8477 ( 2013 ) h. zheng et al , adv . mater . * 21 * , 2175 ( 2011 ) c.g . granqvist , adv . sci . tech . * 55 * , 205 ( 2008 ) j.s.e.m . svensson c.g . granqvist , sol . mater . * 11 * , 29 ( 1984 ) j.j . berzelius , afhandlingar i fysik , kemi och mineralogie * 4 * , 1815 s. k. deb , appl . suppl . * 3 * , 192 ( 1969 ) b.w . faughnan , r.s . crandall , m.a . lampert , appl . lett * 27 * , 275 ( 1975 ) b. reichman , a.j . bard , j. electrochemic . 126 * , 583 ( 1979 ) c.g . granqvist , sol . energ . mat . . cells . * 99 * , 1 - 13 ( 2012 ) j .- g . zhang et al . , `` chromic mechanism in amorphous @xmath5 films '' , 190th electrochemical society meeting , ( 1996 ) m.b .. johansson , g. a. niklasson , l. j. sterlund , j. mater . 27 * , 3130 ( 2012 ) s. raj et al . b * 75 * , 155116 ( 2007 ) r. g. bamwenda , k. sayama , h. arakawa , j. photochem . photobiol . a * 122 * , 175 ( 1999 ) c. g. granqvist , sol . energy mater . cells * 60 * , 201 ( 2000 ) s. tanisaki , j. phys . jpn . * 15 * , 566 ( 1960 ) r. diehl , g. brandt , acta crystallogr . * b 34 * , 1105 ( 1978 ) p. w. woodward , a. w. sleight , t. j. vogt , j. phys . chem . solids * 56 * , 1305 ( 1995 ) c.j howard , v. luca , k.s . knight , j. phys . condens . matter * 14 * , 377 ( 2002 ) b. gerand et al . , * 29 * , 429 ( 1979 ) o. yu . khyzhun , yu . m. solonin , v. d. dobrovolsky , j. alloy . * 320 * , 1 ( 2001 ) k. miyake et al . , j. appl . phys . * 55 * , 2747 ( 1984 ) a. antonaia et al . , electrochimica acta * 46 * , 2221 ( 2001 ) s.s sun , p.h . holloway , j. vac . technol . a. * 2 * , 336 ( 1984 ) k.r . dey et al . , materials research bulletin * 42 * , 591 ( 2007 ) q. zhong , j. r. dahn , k. colbow , phys . b. * 46 * , 2554 ( 1992 ) s - h . lee et al . , solid state ionics * 156 * , 447 ( 2003 ) p.gerard , a. deneuville and r. courths , thin solid films , * 71 * , 221 ( 1980 ) w. wright j. solid state chem . * 20 * , 89 ( 1977 ) p.j . wiseman and p.g . dickens , j . solid state chem . * 6 * , 374 ( 1973 ) p. uppachai et al . , acta * 145 * , 27 ( 2014 ) t. yoshimura , j. appl . 57 * , 911 ( 1985 ) a. deneuville , p. gerard , j. electron . * 7 * , 559 ( 1978 ) o. pyper et al . , sol . energ . mat . sol . c. * 71 * , 511 ( 2002 ) o.f . schirmer et al , j. electrochem . : solid - state science and technology * 124 * , 749 ( 1977 ) e. salje , opt . commun . , * 24 * , 231 ( 1978 ) l. berggren et al . , j. appl . phys . , * 90 * , 1860 ( 2001 ) m. f. saenger et all.,phys . status solidi a , * 4 * , 914 ( 2008 ) s. hashimoto , h. matsuoka , j. appl . phys . * 69 * , 933 ( 1991 ) h. chen et al . , j. appl . * 101 * , 114303 ( 2007 ) m.m . johansson et al . , j. appl . 115 * , 213510 ( 2014 ) g. a. niklarsson et al . , j. of physics : conference series , * 100 * , 082023 ( 2008 ) e. ozkan et al . , sol . energ . mat . sol . c. * 79 * , 439 ( 2003 ) g. leftheriotis et al . , thin solid films * 384 * , 298 ( 2001 ) e. iguchi , h. miyaghi j. phys . chem . solids . * 54 * , 403 ( 1993 ) a. stashans and s. lunell , int . j. quantum chem . * 63 * , 729 ( 1997 ) g. a. de wijs , r.a . de groot , phys . b60 , 16463 ( 1999 ) f. wang et al . , phys . b80 , 144424 ( 2009 ) d. b. migas et . al , j. appl . phys.108 , 093713 ( 2010 ) n.l . heda and b.l . ahuja , comp . mat . science.72 , 49 ( 2013 ) r. chatten et al . , j. phys . chem . * 109 * , 3146 ( 2005 ) c. lambert - mauriata et al . , surface science * 606 * , 40 ( 2012 ) g. kresse and j. hafner , phys . b. * 47 * , 558 , ( 1994 ) , phys . rev . b. * 49 * , 14251 ( 1994 ) g. kresse and j. furthmuler , comput . mat . * 6 * , 15 ( 1996 ) ; phys . b. * 54 * , 11169 ( 1996 ) s. l. dudarev et al . b. * 57 * , 1505 ( 1998 ) j. p. perdew , k. burke , and m. ernzerhof , phys . * 77 * , 3865 ( 1996 ) p. giannozzi et al , j. phys . : condens . matter * 21 * , 395502 ( 2009 ) m. cococcioni and s. de gironcoli , phys . b * 71 * , 035105 ( 2005 ) j. heyd , g. e. scuseria , and m. ernzerhof , j. chem . phys . * 118 * , 8207 ( 2003 ) j. heyd and g. e. scuseria , j. chem . * 120 * , 7274 ( 2004 ) ; n. bondarenko , o. eriksson and n. v. skorodumova , phys . rev . b * 89 * , 125118 ( 2014 ) g. henkelman and h. jonsson j. chem . phys . * 113 * , 9901 ( 2000 ) d. sheppard et al . , j. chem . phys . * 136 * , 74103 ( 2012 ) n. gonzalez - garcia et al , chem . theory comput . * 2 * , 895 ( 2006 ) l. xie et al . , j. chem . phys . * 120 * , 8039 ( 2004 ) s. ong , y. mo and g. ceder , phys . b * 85 * , 081105 ( 2012 ) d. a. andersson et al , phys . rev . b * 75 * , 035109 ( 2007 ) s. arapan , i. simak and n. v. skorodumova , phys . rev . b * 91 * , 125108 ( 2015 ) t. maxisch , f. zhou and g. ceder , phys . rev . b. * 73 * , 104301 ( 2006 ) a. droghetti , c. d. pemmaraju , and s. sanvito , phys . b. * 81 * , 092403 ( 2010 ) v. i. anisimov , j. zaanen , and o. k. andersen , phys . b * 44 * , 943 ( 1991 ) n. deskins , m , dupuis , j. phys . chem . * 113 * , 346 ( 2009 ) x. ma , y. wu , y. lv and y. zhu , j. chem . phys . * 117 * , 26029 ( 2013 ) p. erhart et al . b * 90 * , 035204 ( 2014 ) a. d. becke , j. chem . phys . * 98 * , 5648 ( 1993 ) c. franchini , g. kresse and r. podloucky , phys . lett * 102 * , 256402 ( 2009 ) a.l . larsson et al . solid state ionics * 165 * , 35 ( 2003 )	electron localization and polaron mobility in oxygen deficient as well as li doped monoclinic tungsten trioxide have been studied . we show that small polarons formed in the presence of oxygen vacancy prefer the bipolaronic @xmath0 configuration whereas the @xmath1 configuration is found to be metastable . our calculations suggest that bipolarons are tightly bound by the vacancy and therefore largely immobile . on the contrary , polarons formed as a result of li intercalation can be mobile , the activation energy for polaron jumping in this case varies between 98 and 124 mev depending on the crystallographic direction . the formation of @xmath0 bipolarons in @xmath2 is possible . when situated along @xmath3 $ ] the bipolaronic configuration is 8 mev lower in energy than two separate @xmath4 polarons .
electromagnetic duality in yang - mills - higgs theories was proposed in the work of goddard , nuyts , and olive ( gno ) @xcite . in this work they consider a theory with gauge group @xmath7 and a scalar field @xmath8 in an arbitrary representation which spontaneously breaks @xmath7 to @xmath9 in such a way that @xmath10 is nontrivial , which is the necessary condition for the existence of topological monopoles . since then , the monopole s solutions@xcite and the electromagnetic duality conjecture@xcite were studied intensely in the particular case where the scalar field was in the adjoint representation . in this case , the unbroken gauge group @xmath9 necessarily has a @xmath11 factor which is generated by the scalar field vacuum solution . this @xmath11 factor guarantees that @xmath12 , in which case the theory can have monopole solutions which are a generalization of t hooft - polyakov solution . we shall call them @xmath13 monopoles . the bogomolnyi - prasad_sommerfield ( bps ) @xmath13 monopoles are conjectured to be dual to particles in the adjoint representation in the dual theory for suitable supersymmetric theories@xcite . on the other hand , much less is known in the cases where @xmath8 is not in the adjoint representation and @xmath9 is semisimple . in these cases , since @xmath10 is a cyclic group @xmath0 or a product of cyclic groups , the monopoles are called @xmath0 monopoles . some general properties of these @xmath0 monopoles were analyzed in @xcite , and in @xcite it was proposed that the @xmath1 monopole in @xmath14 @xmath2 super yang - mills should satisfy a duality transformation which is alternative to the gno conjecture . one of the main motivations for the study of monopoles and electromagnetic dualities is their possible application to the problem of confinement in qcd . following the ideas of t hooft and mandelstam , the confinement of particles in qcd must be a phenomenon dual to confinement of monopoles in a superconductor . in this model , the formation of chromoelectric flux tubes in qcd must be due to a monopole condensate . however , it is not clear yet if this condensate is made of @xmath13 monopoles , @xmath0 monopoles , or dirac monopoles . there are some lattice results which indicate that confinement could be related to @xmath0 monopoles condensates@xcite . another more recent application of gno duality is in the geometric langlands program @xcite . in the last years , the ideas of t hooft and mandelstam were applied to supersymmetric non - abelian theories which satisfy electromagnetic duality with @xmath13 monopoles . in particular in @xcite the confinement of @xmath13 monopoles by the formation of magnetic flux tubes or @xmath0 strings in soft broken @xmath15 super yang - mills theories with an arbitrary simple gauge group was analyzed . it was shown that the tensions of these @xmath0 strings satisfy the casimir scaling law in the bps limit , which is believed to be the behavior that the chromoelectric flux tubes in qcd must satisfy . this result indicates that these @xmath0 strings can be dual to qcd chomoelectric strings . in order to understand better the properties of the @xmath0 monopoles , in the present work we obtain explicitly the asymptotic form of the @xmath1 monopoles in @xmath2 yang - mills - higgs theories with the gauge group broken to @xmath3 by a scalar in the @xmath4 representation of @xmath2 or its symmetric part . in this case , one could in principle embed the theory in a ( deformed ) @xmath16 @xmath2 super yang - mills with a hypermultiplet in the @xmath4 representation which has vanishing @xmath17 function , similarly to @xmath15 super yang - mills theory . in the work of gno , they obtained that the possible magnetic weights of the @xmath0 monopoles must belong to the weight lattice @xmath18 of the dual unbroken gauge group @xmath19 , where @xmath20 means the covering group of @xmath9 . the magnetic weights satisfy a further constraint to belong to particular cosets in @xmath18 . in order to obtain that constraint we used the fact that for a theory with an unbroken gauge group @xmath21 , the @xmath0 monopole s topological charge sectors are associated to the elements of the group @xmath22 which is a subgroup of @xmath23 , the center of the group @xmath20 . then , we used the result that the elements of @xmath23 are associated to cosets which are related to nodes of the extended dynkin diagram of @xmath19 related to the node @xmath24 by a symmetry transformation . therefore , the @xmath0 monopoles must be associated to weights of a subset of these cosets . this form of writing elements of the center of a group was also used to obtain @xmath0 strings solutions@xcite . then , using the elegant general construction of weinberg et al.@xcite for the monopoles in theories where the gauge group @xmath2 is broken to @xmath3 , we associated to each weight of the defining representation of the dual algebra @xmath5 a @xmath25 subalgebra and constructed explicitly a @xmath1 monopole asymptotic solution which we called fundamental , generalizing the @xmath1 monopole solution for the @xmath26 gauge group @xcite . this is consistent with the result in @xcite where the authors concluded that the @xmath1 monopoles in these theories should be associated to weights in the coset with the highest weight of the defining representation of @xmath5 . from these fundamental @xmath1 monopoles we constructed other @xmath1 monopole asymptotic solutions . differently from weinberg s general construction where the @xmath0 monopole s topological charge sectors were associated to integers modulo @xmath27 , in our construction they are associated to cosets in the weight lattice @xmath18 , which gives in principle a larger number of possible monopoles . we construct the monopole solutions considering two symmetry breakings of @xmath28 to @xmath29 : one in which @xmath29 is invariant under outer automorphism and another in which it is invariant under cartan automorphism . in the first case , the monopole s magnetic flux is in the cartan subalgebra of @xmath28 but @xmath27 must be odd , and in the second case the magnetic flux is not in the cartan subalgebra of @xmath28 . this general procedure can be generalized to other gauge groups . we expect that this explicit construction of the @xmath1 monopole asymptotic solutions can be useful in order to understand better the electromagnetic duality in the theories where the scalar field is not in the adjoint representation . this paper is organized as follows : we start in sec . 2 giving some mathematical conventions . then , we give a brief review of the results of gno in sec . 3 and explain in sec . 4 how the topological charge sectors of the @xmath0 monopoles are associated to particular cosets in the weight lattice @xmath18 . in sec . 5 , we obtain two scalar field configurations which break @xmath28 to @xmath29 where for the first configuration @xmath29 is invariant under cartan automorphism and for the second configuration it is invariant under outer automorphism . finally , in sec . 6 we construct the fundamental @xmath1 monopoles for both symmetry breaks . we also include an appendix where we analyze the center elements of @xmath30 . let us start by giving some conventions which will be used later on . let @xmath31 be the lie algebra of rank @xmath32 associated to the group @xmath7 . let us adopt the cartan - weyl basis . in this basis , the commutation relations read @xmath33 & = & \left(\alpha\right)^{i}e_{\alpha},\label{3.1}\\ \left[e_{\alpha},e_{-\alpha}\right ] & = & \frac{2\alpha}{\alpha^{2}}\cdot h,\nonumber \end{aligned}\ ] ] where generators @xmath34 , form a basis for the cartan subalgebra ( csa ) @xmath35 , @xmath36 are roots and the upper index in @xmath37 means the component @xmath38 of @xmath36 . given a representation of @xmath31 , we can take a basis \{@xmath39 } in which the elements of the csa , @xmath34 , are diagonal,@xmath40 the vector @xmath41 with the @xmath32 eigenvalues @xmath42 as components is called weight and @xmath39 is called weight state . we shall denote by @xmath43 , the simple roots of @xmath31 which is a basis of the root space and by @xmath44 , @xmath45 , the fundamental weights of @xmath31 . moreover we shall call @xmath46 the simple coroots and fundamental coweights respectively . they satisfy the relations@xmath47 @xmath48 and @xmath49 are respectively simple roots and fundamental weights of the dual algebra @xmath50 . let us denote by @xmath51 the covering group of @xmath7 . then , the fundamental weights form a basis for the weight lattice of @xmath51,@xmath52 this lattice includes as a subset , the root lattice of @xmath7 , @xmath53 which has the simple roots @xmath54 as the basis . similarly , the fundamental coweights @xmath49 are the basis of the weight lattice of the dual group @xmath55@xmath56 which is also called the coweight lattice of @xmath51 and which has the root lattice of the dual group @xmath57(or coroot lattice of @xmath7)@xmath58 as a subset . in a theory with gauge group @xmath7 spontaneously broken to @xmath9 , the monopole s solutions are associated to elements of the second homotopy group@xmath59 this result implies that monopoles are associated with nontrivial elements of @xmath60 which correspond to trivial elements of @xmath61 . therefore , the relation ( [ 3.01 ] ) is equivalent to @xcite@xcite @xmath62 where @xmath63 is the unbroken subgroup of @xmath51 . therefore for simplicity , without loss of generality , we shall consider that @xmath7 is simply connected . let us therefore consider a yang - mills - higgs theory with gauge group @xmath7 which we shall consider to be simple and simply connected . let us also consider that in the theory there is a scalar @xmath8 in a representation @xmath64 of @xmath7 and @xmath65 is a vacuum configuration which spontaneously breaks @xmath7 to a subgroup @xmath9 , in such a way that @xmath10 is nontrivial , which allows the existence of magnetic monopoles . the generators of @xmath9 are those which annihilate @xmath65 , that is,@xmath66 we shall denote by @xmath67 the algebra formed by these generators . let us briefly review some general properties of these monopoles which will also be useful to fix our notation . following gno @xcite , we shall consider static finite energy monopoles with the asymptotic form of the magnetic field of the form@xmath68 where @xmath69 and @xmath70 are the angular spherical coordinates . the finite energy asymptotic condition @xmath71 , implies that asymptotically we can write @xcite@xmath72 where @xmath73 . then , the condition that @xmath74 implies that @xmath75 with @xmath76 . the asymptotic condition @xmath71 and the definition of the field strength as the commutator of covariant derivatives , implies that asymptotically @xmath77 . then , using ( [ eq:3.1 ] ) , ( [ 5.0.2 ] ) , and ( [ eq:3.1d ] ) results that @xmath78 and therefore , @xmath79 . moreover , one can write @xcite @xmath80 where @xmath81 , with @xmath82 being the elements of the cartan subalgebra ( csa ) of @xmath67 and @xmath83 is a constant vector . note that in general , the elements of the csa of @xmath67 do not necessarily belong to the csa of @xmath31 , the lie algebra of @xmath7 . therefore , we shall denote by @xmath84 the generators of @xmath67 and @xmath85 and @xmath86 the generators of @xmath31 . one can show the quantization condition @xcite @xmath87=\exp\left[ie\omega\cdot h\right]=\mathbbm{1}.\label{3.32}\ ] ] let us consider that @xmath9 is semisimple is in the adjoint representation , then @xmath9 is not semisimple since it has a @xmath11 factor generated by the vacuum configuration . this case is considered in detail in @xcite . ] . in this case @xmath9 can be written as@xmath88 where @xmath20 is the universal covering group of @xmath9 and the factor @xmath89 is the kernel of the homomorphism @xmath90 . the factor @xmath89 is a discrete subgroup of the center of @xmath20 which we will denote by @xmath91 therefore , the topological charge sectors of the theory are associated to @xmath92 since @xmath89 is a cyclic group @xmath0 or a product of cyclic groups , then these monopoles are called @xmath0 monopoles . considering the condition ( [ 3.32 ] ) in @xmath20 rather than in @xmath9 implies that @xcite@xmath93\in k(g_{0})\subset z(\widetilde{g}_{0}),\label{eq:3.5}\ ] ] where @xmath94 denotes the exponential mapping in @xmath20 . using the fact that the elements of the center @xmath23 of a group @xmath9 , have the form@xmath95\label{3.4}\ ] ] where @xmath96 is a vector of the coweight lattice @xmath97 , goddard , nuyts , and olive concluded that the so - called magnetic weights must satisfy @xmath98 together with condition ( [ eq:3.5 ] ) . from this result they conjectured that the monopoles should be dual to particles in a theory with unbroken gauge group @xmath19 . let us now analyze how the different values of @xmath99 are associated to the different elements of @xmath100 or topological charge sectors ( [ 3.31 ] ) of the theory . we shall also restrict the possible values @xmath99 can take . in order to do this we must remember that since @xmath101 is a sublattice of @xmath97 , we can define the quotient @xmath102 . this quotient can be represented by the cosets @xcite@xmath103 where the weights @xmath104 are associated to nodes of an extended dynkin diagram of @xmath105 related to the node @xmath24 by a symmetry transformation . in table 1 we used black nodes to denote these nodes in the extended dynkin diagrams . one can then show that the center of @xmath9 is a discrete group isomorphic to the classes@xcite @xmath106,\,\exp\left[2\pi i\left(\lambda_{\tau(0)}^{\vee}+\lambda_{r}(g_{0}^{\vee})\right)\cdot h\right],\, ... \,,\right.\label{3.65}\\ & & \left.,\, ... \,,\,\exp\left[2\pi i\left(\lambda_{\tau^{n}(0)}^{\vee}+\lambda_{r}(g_{0}^{\vee})\right)\cdot h\right]\right\ } .\nonumber \end{aligned}\ ] ] in other words , all the group elements ( [ 3.4 ] ) with the vector @xmath96 in the same coset of ( [ 3.6 ] ) correspond to the same element of the center @xmath23 . in particular when @xmath96 belongs to @xmath101 , the group elements ( [ 3.4 ] ) correspond to the identity of @xmath23 . such a way of writing the center elements is also quite useful to analyze the @xmath107 string solutions which appear when the gauge group @xmath7 is broken to its center group @xmath108@xcite@xcite@xcite . in @xcite the center group of some groups is analyzed in some more detail . .[cap : extended - dynkin][dynkin]extended dynkin diagrams , nodes symmetrically related to the node @xmath24 and center groups @xmath108 . [ cols="^,^,^,^",options="header " , ] since @xmath100 , it will be formed by a subset of elements of ( [ 3.65 ] ) . therefore , the magnetic weights @xmath99 can not belong to an arbitrary coset of ( [ 3.6 ] ) , but only to those cosets associated to elements of @xmath89 , the kernel of the homomorphism @xmath90 . moreover , @xmath0 monopoles will be in the same topological sector if their associated magnetic weights are in the same coset . in the next sections we will analyze the values the magnetic weights can take when the gauge group @xmath2 is broken to @xmath109 let us now consider a yang - mills theory with gauge group @xmath2 and with a scalar field @xmath8 in the representation which is the direct product @xmath4 of @xmath2 . in this case the theory can in principle be embedded in a ( deformed ) @xmath16 @xmath2 super yang - mills with an hypermultiplet in the @xmath4 representation which has a vanishing @xmath17 function . such a theory has already been considered in @xcite but with a different gauge symmetry breaking , which gave rise to @xmath110 strings and @xmath13 monopole confinement . we can also consider that @xmath8 is in the symmetric part of @xmath4 , since the vacuum solutions @xmath65 which we will consider are nontrivial only in the symmetric part of @xmath4 . the specific form of the potential is not important to determine the asymptotic form of the @xmath1 monopoles . in order for @xmath1 monopoles to exist , we want to find configurations of the scalar field which break @xmath2 to @xmath3 , @xmath111 . we first analyze some @xmath29 subalgebras of @xmath28 . we shall consider that @xmath112 , since @xmath113 is not simple . we shall consider @xmath29 invariant subalgebras under cartan or outer automorphisms which are order two automorphisms or involutions . remembering that , under an order two automorphism @xmath114 of a lie algebra @xmath31 , @xmath114 has eigenvalues @xmath115 and the lie algebra split into @xmath116 and @xmath117 , where @xmath118 is formed by the generators @xmath119 , such that @xmath120 . moreover , @xmath116 forms a subalgebra of @xmath31 . the quotient of the group generated by @xmath31 modulo the group generated by @xmath116 is a symmetric space and @xmath117 is associated to a representation of @xmath116@xcite . the cartan automorphism for a general lie algebra @xmath31 is defined by@xmath121 for this automorphism , @xmath116 and @xmath117 are formed by the generators @xmath122 let us consider @xmath123 . the generators of @xmath28 in the @xmath27-dimensional representation can can be written in terms of the @xmath4 matrices @xmath124 with components @xmath125 or @xmath126 where @xmath127 are the weight states of the @xmath27-dimensional representation . then , the basis elements of the csa of @xmath28 correspond to the traceless combinations @xmath128@xmath129 . on the other hand , the generator @xmath130 is the step operator associated to the root @xmath131 , where @xmath132 are orthonormal vectors in the @xmath27-dimensional vector space with @xmath133 . the root @xmath131 is positive ( negative ) if @xmath134 @xmath135 and is a simple root if @xmath136 the weight associated to @xmath127 can be written as@xmath137 from ( [ 1.1 ] ) , we can conclude that the generators of @xmath116 , in the @xmath27-dimensional representation are@xmath138 which are @xmath139 antisymmetric @xmath4 matrices which form a @xmath29 subalgebra of @xmath28 . therefore , for @xmath140 @xmath141 for @xmath142@xcite . for example , in the @xmath143 case , the gell - mann matrices @xmath144 form a @xmath145 invariant subalgebra under cartan automorphism . let us consider the scalar field configuration@xmath146 where @xmath96 is a constant . using the fact that in a tensor product representation a generator @xmath147 acts as @xmath148 , it is straightforward to conclude that @xmath65 is annihilated only by the generators ( [ 1.6 ] ) and hence _ breaks _ @xmath28 to @xmath29 subalgebra generated by ( [ 1.6 ] ) . therefore , we shall consider that @xmath65 is the vacuum configuration responsible for the symmetry breaking . let us determine an orthogonal basis of the cartan subalgebra ( csa ) of the above @xmath29 invariant subalgebra ( [ 1.6 ] ) . for @xmath145 , we can consider the gell - mann matrix @xmath149 as the generator of the cartan subalgebra . recalling that @xmath150 and @xmath151 have same rank equal to @xmath152 , one can check that the generators @xmath153 form an orthogonal basis of the cartan subalgebras of @xmath29 for @xmath154 , and where @xmath155 are generators of @xmath28 . it is important to note that for this @xmath29 subalgebra invariant under cartan automorphism , the csa of @xmath29 is not in the csa of @xmath28 . we shall denote by @xmath82 , @xmath156 the generators of the subalgebra @xmath29 in order to distinguish from generators @xmath85 , @xmath86 of @xmath28 . a dynkin diagram which is invariant under a transformation of the nodes , @xmath157 , implies that the corresponding cartan matrix satisfies@xmath158 as a consequence the associated lie algebra @xmath31 has an outer automorphism @xmath159:@xmath160 where @xmath161 . in particular , for @xmath162 , the dynkin diagram is invariant under the transformation of the nodes @xmath163 . let @xmath164 denote the orbit of nodes @xmath165 and @xmath166 connected under this transformation . using the so - called folding procedure one can show that the invariant subalgebra of @xmath162 under this automorphism is @xmath151 @xcite . let @xmath85 and @xmath167 be generators of @xmath162 . then , the invariant @xmath151 subalgebra has the following generators@xmath168 where @xmath169 . let @xmath54 and @xmath44 be , respectively , simple roots and fundamental weights of @xmath162 . then , the simple roots , simple coroots , fundamental weights and coweights of the invariant subalgebra @xmath151 are@xmath170 one can check easily that the scalar products between the simple roots give the cartan matrix of @xmath171 and that simple roots and fundamental weights satisfy the right orthonormality conditions . as in the cartan automorphism , we are looking for a vacuum configuration @xmath65 which is annihilated by the generators given by ( [ eq:1.3 ] ) , that is , which breaks @xmath162 to the @xmath151 subalgebra invariant by outer automorphism . let us consider the scalar field configuration @xmath172 where @xmath96 is a constant . since @xmath173 , we can obtain directly that @xmath174 with respect to the folded step operators @xmath175 we can use the fact that @xmath176 which implies that @xmath177 therefore we can consider that @xmath65 given by eq.([4.2.5 ] ) is a configuration which breaks @xmath162 to the @xmath151 subalgebra invariant by outer automorphism . the above @xmath29 subalgebras of @xmath28 generates subgroups @xmath178 of @xmath2 , where @xmath89 is a subgroup of the center of @xmath179 which we want to determine . following @xcite , in order to determine the factor @xmath89 of the subgroup @xmath180 of @xmath181 , we must first choose a representation @xmath182 of @xmath51 with highest weight @xmath183 such that the @xmath184 . if @xmath182 branches to the representation @xmath185 of @xmath20 with highest weight @xmath186 , then @xmath187 . therefore , in order to determine the discrete group @xmath89 of the unbroken gauge subgroup @xmath188 of @xmath2 we shall choose the @xmath27-dimensional representation @xmath189 of @xmath2 with highest weight @xmath190 since in this representation @xmath191 is well known . then , for the above two different embeddings of @xmath29 in @xmath28 , the @xmath27-dimensional irrep . of @xmath28 branches to the @xmath27-dimensional irreducible representation ( irrep ) of @xmath29 , which has @xmath190 as highest weight . the weight states of this representation are of the form @xmath192 , where @xmath193 are positive roots of @xmath29 . the kernel @xmath194 of this representation is made by the elements @xmath195 such that@xmath196 for all weight states @xmath192 of the representation . since @xmath197 is a subgroup of the center @xmath198 , we just need to act the elements of @xmath198 on the weight states of the representation . let us consider @xmath179 with @xmath199 . in the appendix we analyze the particular cases of the groups @xmath179 for @xmath200 . from the symmetry of the extended dynkin diagram in table 1 we can conclude that the weight lattice @xmath201 split in@xmath202 and the center of @xmath203 is@xmath204\right\ } , \ ] ] where @xmath205 . acting these elements on the weight states of the @xmath206-dimensional representation of @xmath207 we obtain@xmath208\left\|\lambda_{1}-\gamma\right\rangle & = & \left\|\lambda_{1}-\gamma\right\rangle .\nonumber \end{aligned}\ ] ] for @xmath209 we have@xmath210 where @xmath211 . in both cases the weight lattice split in the four cosets,@xmath212 and@xmath213,\right.\\ & & \hfill\left.\exp\left[2\pi i\left(\lambda_{n-1}^{\vee}+\alpha^{\vee}\right)\cdot h\right],\exp\left[2\pi i\left(\lambda_{n}^{\vee}+\alpha^{\vee}\right)\cdot h\right]\right\ } , \end{aligned}\ ] ] where @xmath214 . acting these elements on the weight states of the @xmath215-dimensional representation of @xmath216 we obtain@xmath217\left\|\lambda_{1}-\gamma\right\rangle & = & \left\|\lambda_{1}-\gamma\right\rangle , \label{2.312}\\ \exp\left[2\pi i\left(\lambda_{n-1}^{\vee}+\alpha^{\vee}\right)\cdot h\right]\left\|\lambda_{1}-\gamma\right\rangle & = & -\left\|\lambda_{1}-\gamma\right\rangle , \nonumber \\ \exp\left[2\pi i\left(\lambda_{n}^{\vee}+\alpha^{\vee}\right)\cdot h\right]\left\|\lambda_{1}-\gamma\right\rangle & = & -\left\|\lambda_{1}-\gamma\right\rangle .\nonumber \end{aligned}\ ] ] therefore , from ( [ 2.311 ] ) and ( [ 2.312 ] ) , we can conclude that for @xmath27 odd or even , @xmath218,\exp\left[2i\pi\left(\lambda_{1}^{\vee}+\lambda_{r}(spin(n)^{\vee})\right)\cdot h\right]\right\ } \end{aligned}\ ] ] and the symmetry breaking is of the form@xmath219 hence we can conclude that for this symmetry breaking , the magnetic weights @xmath99 must belong to the cosets@xmath220 where @xmath221 is the highest weight of the defining representation of @xmath5 which has dimension @xmath222 for @xmath223 and for @xmath224 . this result holds also for the special cases @xmath225 and @xmath226 , as is analyzed in the appendix . a similar result was obtained in @xcite using a different approach . we want to construct explicitly the asymptotic form for static spherically symmetric @xmath1 monopole s solutions . in order to do that let us define@xmath227 where@xmath228 and from eq . ( [ eq:3.5 ] ) implies that@xmath229\in k(g_{0}).\ ] ] let us also consider other two generators @xmath230 , @xmath231 of @xmath31 , but not of @xmath67 , such that @xmath232=i\epsilon_{ijk}t_{k}^{\beta}.\ ] ] the choice of these two generators @xmath230 and @xmath231 will be discussed in detail in sections [ sub : so(n)-invariant - under ] and [ sub : so(2m+1)-invariant - under ] . following e. weinberg et al . @xcite , we can construct a spherically symmetric monopole , consistent with gno results , with the asymptotic form of the scalar field given by ( [ 5.0.2 ] ) with the group element @xmath233 of the form @xmath234\exp[-i\theta t_{2}^{\beta}]\exp[i\varphi t_{3}^{\beta}]\ ] ] and the asymptotic form for the gauge field given by @xcite@xmath235 where@xmath236 this gauge field produces the magnetic field @xmath237 consistent with eqs . ( [ eq:3.1 ] ) , ( [ eq:3.1d ] ) . note that in our construction , we have a difference from weinberg s construction . in his construction , the @xmath110 monopoles in a given topological charge class were associated to the same integer modulo @xmath238 . on the other hand , in our construction , @xmath110 monopoles are in the same topological sector when they are associated to magnetic weights @xmath239 in the same coset in ( [ 3.6 ] ) . using the identity@xmath240 where @xmath241 is a constant and @xmath242 , @xmath243 are generators of a @xmath25 algebra , we can write the asymptotic form of the gauge fields ( [ 5.0.1 ] ) as@xmath244,\\ w_{\varphi}(\theta,\varphi ) & = & \frac{\sin\theta}{e}\left[-\left(\sin\theta\right)t_{3}^{\beta}+\cos\theta\left(\left(\cos\varphi\right)t_{1}^{\beta}+\left(\sin\varphi\right)t_{2}^{\beta}\right)\right],\\ w_{r}(\theta,\varphi ) & = & 0.\end{aligned}\ ] ] using eq . ( [ 5.0.3 ] ) we can also rewrite the @xmath1 monopole asymptotic magnetic field ( [ 5.0.4 ] ) as @xmath245\nonumber \\ & = & \frac{r_{i}}{er^{4}}\sum_{j=1}^{3}r_{j}t_{j}^{\beta},\label{5.0.6}\end{aligned}\ ] ] which is the standard hedgehog form for the magnetic field . let us now analyze the possible monopole solutions for both symmetry breakings discussed in the previous sections . let us determine the possible @xmath25 subalgebra s generators @xmath246 for the symmetry breaking of @xmath247 where @xmath179 is the subgroup invariant under cartan automorphism . we have that @xmath248 where the cartan elements @xmath82 are given by ( [ eq:1.2a ] ) . then , from eq . ( [ 4.3.5 ] ) , we conclude that for the @xmath1 monopoles associated to the nontrivial sector , the vector @xmath17 must belong to the coset @xmath249 this coset has in particular the weights of the defining representation of the dual algebra @xmath5 which has @xmath221 as the highest weight . we know that @xmath223 and @xmath224 , and that the weights of the defining representation of @xmath150 and @xmath250 have dimension @xmath222 . in terms of the orthonormal vectors these weights can be written in both cases as @xmath251 for each weight @xmath252 , we can construct a @xmath25 subalgebra@xmath253 for @xmath254 . from eq . ( [ 1.1 ] ) we can conclude that @xmath255 and @xmath256 . therefore , for each weight of the defining representation of @xmath5 we have a @xmath1 monopole solution ( [ 5.0.2 ] ) , ( [ 5.0.1]),([5.0.6 ] ) . we can construct monopole asymptotic forms with magnetic charge associated to others elements of the cosets ( [ 4.3.5 ] ) . however , the @xmath25 generators associated to these new monopoles seem to be always combination of the generators ( [ 6.1.1 ] ) and therefore these monopoles can be interpreted as superpositions of the above monopoles which we call fundamental . some examples of these @xmath25 subalgebras are @xmath257 for @xmath258 . we can understand easily the @xmath1 nature of these monopoles : writing the fundamental weight @xmath190 in the basis of simple roots , we have that for @xmath259,@xmath260 and for @xmath261,@xmath262 in both cases we see that @xmath263 , and therefore a combination of an even number of fundamental @xmath1 monopoles will result in a configuration associated to @xmath264 which corresponds to the trivial element @xmath265 of the group @xmath1 , and an odd combination will result on a configuration associated to the nontrivial coset , which corresponds to element @xmath266 of the group @xmath1 . we must also note that two configurations belonging to the same topological sector does not have necessarily the same magnetic charge or magnetic weight . from the vacuum ( [ eq:1.2 ] ) , we obtain that the asymptotic form of the scalar field for the monopole associated to the magnetic weight @xmath267 is@xmath268 where we defined @xmath269 from the kinetic term for the scalar field , expanding @xmath8 around the vacuum @xmath65 , we obtain the term @xmath270 which implies that the mass squared matrix for the gauge particles is@xmath271 for @xmath28 broken to @xmath29 , there are @xmath272 massive gauge particles which can be associated to the generators@xmath273 then , using the definition of @xmath124 and adopting the normalization @xmath274 one can obtain directly that all massive gauge particles have same mass equal to @xmath275 this result coincides with the one obtained in @xcite for the @xmath26 case , up to a global factor due to a different normalization . let us determine the possible @xmath25 subalgebra generators @xmath246 we can have for the symmetry breaking of @xmath276 where @xmath277 is the subgroup invariant under outer automorphism . ( [ 4.3.5 ] ) , we can conclude that for the @xmath1 monopoles in the nontrivial sector , the vector @xmath17 must belong to coset @xmath278 where @xmath279 are integer numbers and @xmath280 and @xmath281 are , respectively , coweights and coroots of @xmath151 given by the set of eqs . ( [ 4.2.3 ] ) . from these equations and the fact that @xmath282 where @xmath283 is the highest root of @xmath162 , we can conclude that@xmath284 therefore , @xmath17 must also belong to the subspace of @xmath285 invariant under the outer automorphism transformation @xmath286 of the @xmath162 algebra . the fact that at the same time @xmath287 and @xmath288 is consistent with eq . ( [ 3.01 ] ) which means that the @xmath1 monopoles must be associated to elements of @xmath289 which correspond to the identity of @xmath290 . in order to construct @xmath25 subalgebras we consider that @xmath17 must satisfy not only condition ( [ 5.1.4 ] ) but also that it is a root of @xmath291 then we can define @xmath292 where we used the fact that @xmath293 since it is a root of @xmath162 . since the roots of @xmath162 are of the form@xmath294 where @xmath295 , and @xmath17 must satisfy ( [ 5.1.4 ] ) , we arrive to the conclusion that @xmath17 can be the following @xmath222 roots,@xmath296 where we wrote the roots of @xmath162 as coweights of @xmath151 , using eq . ( [ 4.2.3 ] ) . from the fact that fundamental coweights and simple coroots of @xmath151 are , respectively , fundamental weights and simple roots of @xmath297 we can recognize this set as the weights of the @xmath222-dimensional defining representation of @xmath250 . it remains to show that @xmath298 . in order to do that we can write the set of roots ( [ 5.1.7 ] ) in terms of the orthonormal basis vectors @xmath299 . since @xmath300 , the roots in ( [ 5.1.7 ] ) are of the form @xmath301 consistent with the fact that under the outer automorphism of @xmath302@xmath303 . the step operator associated to the root ( [ 5.1.8 ] ) in the @xmath27-dimensional representation is proportional to the matrix @xmath304therefore , denoting @xmath305 we can write @xmath230 and @xmath231 in terms of these matrices and acting on the vacuum ( [ 4.2.5 ] ) we obtain@xmath306 @xmath307 hence , @xmath298 . therefore , we can conclude that to each weight ( [ 5.1.7 ] ) of the defining representation of @xmath259 we can associate a @xmath25 subalgebra ( [ 5.1.5 ] ) and @xmath1 monopole . similarly to the previous case , we can construct monopole asymptotic forms with magnetic charge associated to others elements of the cosets ( [ 4.3.5 ] ) . however , the @xmath25 generators associated to these new monopoles seem to always be combination of the generators ( [ 5.1.5 ] ) and therefore these monopoles can be interpreted as superpositions of the above monopoles . some examples of these @xmath25 subalgebras are generated by@xmath308 where @xmath309 . for the vacuum ( [ 4.2.5 ] ) , we obtain that the asymptotic form of the scalar field for the monopole associated to @xmath310 is@xmath311+\right.\\ & & + \left.\left(\cos\theta-1\right)\left[\left\|p,2m+2-p\right\rangle + \left\|2m+2-p , p\right\rangle \right]\right\ } .\end{aligned}\ ] ] let us analyze the elements of the center group and the kernel @xmath89 for the special cases of @xmath179 for @xmath200 : for @xmath312 , the weight lattice splits in two cosets,@xmath313 and@xmath314\right\ } .\ ] ] the branching of the irrep . with highest weight @xmath190 of @xmath143 in @xmath25 is @xmath315 . acting the center elements on the weight states @xmath316 of @xmath25 , we obtain that@xmath317,\exp\left[2\pi i\left(\lambda_{1}^{\vee}+\lambda_{r}(su(2))\right)\cdot h\right]\right\ } , \ ] ] where @xmath221 is the highest weight of the @xmath318-dimensional irrep . of @xmath25 . for @xmath319 , the weight lattice splits in two cosets,@xmath320 and@xmath321\right\ } .\ ] ] the branching of the irrep . with highest weight @xmath190 of @xmath322 in @xmath323 is @xmath324 . acting the center elements on the weight states @xmath325 of @xmath323 , we obtain that@xmath326,\exp\left[2\pi i\left(\lambda_{2}^{\vee}+\lambda_{r}(sp(4)^{\vee})\right)\cdot h\right]\right\}\ ] ] where @xmath327 is the highest weight of the @xmath328-dimensional irrep . of @xmath323 . for @xmath329 , the weight lattice splits in four cosets , @xmath330 and@xmath331,\right.\\ & & \hfill\left.\exp\left[2\pi i\left(\lambda_{2}^{\vee}+\alpha^{\vee}\right)\cdot h\right],\exp\left[2\pi i\left(\lambda_{3}^{\vee}+\alpha^{\vee}\right)\cdot h\right]\right\ } .\end{aligned}\ ] ] the branching of the irrep . with highest weight @xmath190 of @xmath332 in @xmath333 is @xmath334 . acting the center elements on the weight states @xmath325 of @xmath333 , we obtain that@xmath335,\exp\left[2\pi i\left(\lambda_{2}^{\vee}+\lambda_{r}(su(4))\right)\cdot h\right]\right\}\ ] ] where @xmath327 is the highest weight of the @xmath336-dimensional irrep . of @xmath337 . therefore for all the three cases , similarly to the general case of @xmath338 , @xmath339,\exp\left[2\pi i\left(\lambda_{i}^{\vee}+\lambda_{r}(spin(n)^{\vee})\right)\cdot h\right]\right\ } , \ ] ] where @xmath49 is the highest weight of the defining representation of @xmath5 which has dimension @xmath222 for @xmath223 and for @xmath224 . 10 p. goddard , j. nuyts and d.i . olive , nucl . b125 , 1 ( 1977 ) . p. goddard and d.i . olive , rept.prog . 41 , 1357 ( 1978 ) . bais , phys . rev d18 , 1206 ( 1978 ) . weinberg , phys . d20 , 936 ( 1979 ) ; nucl . b167 , 500 ( 1980 ) ; nucl . b203 , 445 ( 1982 ) . p. goddard and d.i . olive , nucl . b191 , 511 ( 1981 ) ; nucl . b191 , 528 ( 1981 ) . c.montonen and d.i.olive , phys . b72 , 117 ( 1977 ) ; e. witten and d.i . olive phys . lett b78 , 97 ( 1978 ) ; h. osborn , phys . lett . b83 , 321 , ( 1979 ) ; a. sen , phys . b329 , 217 ( 1994 ) ; n. seiberg and e. witten , nucl . b426 , 19 ( 1994 ) ; nucl . b431 , 484 ( 1994 ) ; c. vafa and e. witten , nucl . b431 , 3 , ( 1994 ) . weinberg , d. london and j.l . rosner , nucl . b236 , 90 ( 1984 ) . f.a . bais and r. laterveer , nucl . b307 , 487 ( 1988 ) . m. strassler , jhep 9809 , 017 ( 1998 ) . i.g . halliday and a. schwimmer , phys . b101 , 327 ( 1981 ) ; phys . b102 , 337 ( 1981 ) . a. kapustin , e. witten , comm . number theory phys . 1 , 1 ( 2007 ) . kneipp , phys . rev . d 69 , 045007 ( 2004 ) . olive , n. turok and j.w.r . underwood , nucl . b409 , 509 ( 1993 ) . kneipp , phys . d76 , 125010 ( 2007 ) . kneipp , jhep 0811 , 049 ( 2008 ) . s. weinberg , `` the quantum theory of fields '' ( cambridge university press , cambridge , england , 1996 ) , vol . 2 , chap . 23 . olive and n. turok , nucl . b215 , 470 ( 1983 ) . kneipp and p. brockill , phys . d64 , 125012 ( 2001 ) ; m.a.c . kneipp , phys . d68 , 045009(2003 ) . s. helgason , `` differential geometry , lie groups , and symmetric spaces '' ( academic press , san diego , usa , 1978 ) . j. cornwell , `` group theory in physics '' ( academic press , san diego , usa , 1984 ) , vol . 2 , chap . 18 . d. london , phys . d 33 , 3075 ( 1986 ) .	@xmath0 monopoles are important for the understanding of the goddard - nuyts - olive duality when the scalar field is not in the adjoint representation . we analyze the @xmath1 monopole solutions in @xmath2 yang - mills - higgs theories spontaneously broken to @xmath3 by a scalar in the @xmath4 representation . we construct explicitly a @xmath1 monopole asymptotic form for each of the weights of the defining representation of the dual algebra @xmath5 . @xmath6 * @xmath1 monopoles in @xmath2 yang - mills - higgs theories * * * marco a.c . kneipp * * * * and paulo j. liebgott * * _ universidade federal de santa catarina ( ufsc ) , + departamento de fsica , cfm , + campus universitrio , trindade , + 88040 - 900 , florianpols , brazil . + _
magnetite is an archetypal strongly correlated transition metal oxide , with properties not well described by single - particle band structure . below 858 k , magnetite , which may be written as fe@xmath2(fe@xmath3fe@xmath4)@xmath5o@xmath1 , is ferrimagnetically ordered , with the a and b sublattices having oppositely directed magnetizations . the moments of the five unpaired @xmath6 electrons of the tetrahedrally coordinated a - site fe@xmath4 ions are compensated by those of the octahedrally coordinated b - site fe@xmath4 ions . the net magnetization results from the octahedrally coordinated b - site fe@xmath3 that have four unpaired @xmath6 electrons @xcite . upon cooling , bulk magnetite undergoes a first - order phase transition from a moderately conducting high temperature state to a more insulating low temperature state at what is now called the verwey@xcite temperature , @xmath7 k. the change in electronic properties is coincident with a structural transition from a high temperature cubic inverse spinel to a low temperature monoclinic unit cell . the nature of the ordered insulating state remains an active topic of current research@xcite . experiments indicate the onset of multiferroicity@xcite in magnetite below 40 k@xcite , further highlighting the rich physics in this correlated system . recently , nanostructured electrodes have been used to apply strong electric fields in the plane of magnetite films@xcite . below @xmath8 , a sufficient applied voltage triggers a breakdown of the comparatively insulating low - temperature state and a sudden increase in conduction @xcite . this is an example of electric field - driven breakdown of a gapped state in strongly correlated oxides@xcite similar to landau - zener breakdown in classic semiconductors . the electric field - driven transition in magnetite is consistent with expectations@xcite based on such a mechanism ( via geometric scaling @xcite , lack of intrinsic hysteresis @xcite , changes of both contact and bulk resistance at the transition @xcite ) . these prior experiments examined films of various thicknesses , from 30 nm to 100 nm . no strong thickness dependence was observed in the switching properties , consistent with the applied lateral electric field at the sample surface acting as the driver of the breakdown ( though thinner films showed a less pronounced verwey transition in low - bias resistance vs. temperature measurements , consistent with expectations ) . here we report studies of the statistical variations of this electric field - driven transition in fe@xmath0o@xmath1 , as a function of temperature and magnetic field perpendicular to the film surface ( out - of - plane ) . we find that there is a statistical distribution of switching voltages , @xmath9 , that becomes more broad and shifts to higher voltages as @xmath10 is reduced . we discuss these trends in the context of switching kinetics in other systems that exhibit similar trends . the application of a magnetic field perpendicular to the plane of the fe@xmath0o@xmath1 film alters the @xmath9 , shifting the mean by several mv ( several percent ) and changing its shape , within a range of fields comparable to that required to reorient the magnetization out of plane . distribution experiment . ( a and b ) the fragments of @xmath11-@xmath12 curves in the vicinity of a transition demonstrating one ( a ) and three ( b ) switching events in a single pulsed @xmath11-@xmath12 cycle . ( c ) typical sem image of ti / au electrodes , patterned on magnetite film surface , separated by nanogap @xmath13 100 nm . ( d ) an example of @xmath9 distribution histogram at 90 k.,width=377 ] the 50 nm fe@xmath0o@xmath1 ( 100 ) thin films used in the present study were grown on ( 100 ) oriented mgo single crystal substrates as described elsewhere @xcite . contact electrodes ( 2 nm adhesion layer of ti and 15 nm layer of au ) were patterned by e - beam lithography on the surface of the fe@xmath0o@xmath1 film . as before @xcite , @xmath9 scales linearly with the channel length , @xmath14 ( the electrode spacing ) , implying an electric field - driven transition . long channels ( @xmath15 100 nm ) required large switching voltages that would alter the electrode geometry over numerous switching cycles , distorting the shape of @xmath9 histograms . to minimize @xmath9 , electrodes separated by 10 - 30 nm were patterned using a self - aligned technique @xcite . electrical characterization of the devices was performed using a semiconductor parameter analyzer ( hp 4155a ) . to minimize self - heating when in the conducting state , the voltage was applied as pulses 500 @xmath16s in duration with a 5 ms period @xcite . the samples were cooled below @xmath8 with no magnetic field applied , and the distribution of @xmath9 was obtained by executing several thousand consecutive forward pulsed @xmath11-@xmath12 sweeps in the vicinity of the transition point ( typically a 0.2 - 0.3 v range ) at a fixed ( to within 50 mk ) temperature , and recording the number of switching events at each voltage . each voltage value is essentially an independent test to see if switching takes place under the pulse conditions . hence , some sweeps show one ( figure [ fig1 ] a ) or several ( figure [ fig1 ] b ) switching events . even if the system is switched to the conducting state at @xmath9(1 ) , it may return to the off state between pulses , and then switch to the on state at some higher voltage , @xmath9(2 ) , and so on . the @xmath9 distribution at a particular temperature is built by recording all switching events over several thousands ( 3000 - 6000 ) of @xmath11-@xmath12 cycles and then counting the number of switchings at a certain @xmath9 , to produce a `` # of counts '' vs. @xmath9 histogram . a typical @xmath9 distribution at 90k is shown in figure [ fig1 ] d. the distribution is a single peak , symmetrical around the most probable @xmath9 value . this procedure was repeated at each temperature below @xmath8 ( @xmath17 110k for devices under test ; see figure [ fig2]b inset ) , down to @xmath17 75 k. at @xmath18 k , the high values of @xmath9 necessary led to irreversible alteration of the electrode geometry , resulting in asymmetric , distorted @xmath9 histograms . near 80 - 90 k , @xmath11-@xmath12 cycles with multiple @xmath9 events ( figure [ fig1]b ) were observed more frequently . thus , the total number of switching events observed varied with @xmath10 , even with a fixed number of @xmath11-@xmath12 cycles at each temperature . to compare @xmath9 distributions at different temperatures , the distributions were normalized , plotted as ( # of counts)/(max # of counts ) vs @xmath9 , where `` max # of counts '' is the number of events at the most probable @xmath9 and `` # of counts '' is the number of events at a certain @xmath9 . figure [ fig2]a is an example of normalized @xmath9 distributions in the 77 k-105 k temperature range . the measured widths of the @xmath9 distributions are not limited by temperature stability . as has been discussed elsewhere@xcite , the use of pulses is essential to minimize the role of self - heating once the system has been driven into the more conducting state . this self - heating and the short timescale@xcite required to raise the local temperature in the channel significantly makes it extremely challenging to determine directly whether the initial breakdown takesplace through the formation of a conducting filament or through a uniform switching ; once a highly conducting path is formed , the whole channel rapidly becomes conducting through self - heating . the filamentary picture is certainly likely , based on other breakdown phenomena in solids , and the statistical variation in @xmath9 is consistent with the idea of a process involving run - to - run variability associated with _ details rather than global material properties , but this is not definitive . distributions at different temperatures ( 77 k - 105 k ) . ( b ) temperature dependence of the mean switching voltage , @xmath19 . inset shows zero - bias @xmath20 vs @xmath10 plot demonstrating @xmath21 110 k.(c ) temperature dependence of @xmath9 distribution width , @xmath22 ( black circles ) . , width=377 ] the @xmath9 distribution at each temperature is characterized by two main parameters : the mean switching value @xmath23 , where @xmath24 is the total number of switching events ; and the width of distribution , calculated as a standard deviation : @xmath25 . as expected , the @xmath26 has the same @xmath10-dependence ( see figure [ fig2 ] b ) as @xmath27 in single @xmath11-@xmath12 experiments described in previous publications @xcite . more interesting is the @xmath22 temperature dependence , showing broadening of the @xmath9 distribution as the temperature decreases ( figure [ fig2 ] c ) . we note a deviation from monotonous temperature dependence of @xmath22 at 100 k , observed in several devices tested . this is a temperature well below @xmath28 ( see figure [ fig2]b inset ) , where several physical parameters ( resistance , heat capacity and magnetoresistance ) change abruptly . this increase in @xmath22 as temperature decreases is rather counter - intuitive . one might expect `` freezing '' of temperature fluctuations and decrease in thermal noise the temperature decreases and , thus , narrowing of @xmath9 distributions . the field - driven breakdown of the insulating state is an example of the `` escape - over - barrier '' problem , addressed generally by a. garg @xcite . below @xmath8 , the ( temperature - dependent ) effective free energy of the electronic system is at a global minimum value in the insulating state , while the external electric field modifies the free energy landscape , lowering the free energy of another local minimum corresponding to the more conducting state . as the external field is increased beyond some critical value , the minimum corresponding to the more conducting state becomes the global minimum . the nonequilibrium transition to the conducting state then corresponds to some process that crosses the free energy barrier between these minima . at a sufficiently large value of the external field , the free energy has only one minimum , corresponding to the conducting state . as garg showed , one may consider thermal activation over the free energy barrier as well as the possibility of quantum escape . this free energy picture predicts a broadening of the transition driving force ( @xmath9 here ) distribution as the absolute value of the driving force increases , consistent with our observations . this free energy picture has proven useful in studying other nonequilibrium transitions , such as magnetization reversal in nanoparticles @xcite and nanowires @xcite . pinning due to local disorder is one way to find increasing distribution widths as @xmath29 , as seen in investigations of field - driven magnetization reversal in nanowires @xcite . unfortunately , quantitative modeling in this framework requires several free parameters and is difficult without a detailed understanding of the underlying mechanism . qualitatively similar phenomenology ( distribution of switching thresholds that broadens as @xmath10 is decreased ) is also observed in the current - driven superconducting - normal transition in ultrathin nanowires@xcite . in this latter case as in ours , self - heating in the switched state is of critical importance , as is the temperature variation of the local thermal path . again , quantitative modeling using this self - heating approach would require the introduction of multiple parameters that are difficult to constrain experimentally , as well as detailed thermal modeling of the nanoscale local effective temperature distribution , and is beyond the scope of this paper . we also examined the dependence of the switching distributions on applied out - of - plane magnetic field , @xmath30 . @xmath9 distributions ( 3000 cycles each ) were collected consecutively at 12 magnetic field values : 0 t ( first ) @xmath31 0.2 t @xmath31 0.4 t @xmath31 0.6 t @xmath31 0.8 t @xmath31 1 t @xmath31 2 t @xmath31 3 t @xmath31 4 t @xmath31 5 t @xmath31 6 t @xmath31 0 t(last ) . figure [ vsw_h_dep ] a shows the resultant @xmath9 distributions at 80 k at several selected magnetic fields . as can be seen , magnetic field shifts the @xmath9 peak to higher @xmath12 values and narrows the @xmath9 distributions . to reassure that the observed @xmath32 shift is not from irreversible changes in the device , a control experiment returning to @xmath33 t was performed after experiments in all non - zero magnetic fields . the @xmath9 distributions at @xmath33 t initially [ @xmath33 t ( first ) ] and in the end [ @xmath33 t ( last ) ] are identical ( see figure [ vsw_h_dep ] a ) , meaning that observed changes in @xmath9 distributions ( shift and narrowing ) are indeed caused by the applied magnetic field . figure [ vsw_h_dep ] b quantifies the dependence of @xmath32 and @xmath22 . it is clear that both parameters saturate as @xmath30 is increased beyond 1 t , _ i.e. _ ; further increases of @xmath30 up to 6 t have no significant effect . magnetic field of the opposite polarity ( not shown ) has exactly the same effect on the position and the width of @xmath9 distributions . note that the shape of the distribution , in particular its asymmetry about the peak value of @xmath9 , evolves nontrivially with magnetic field , becoming more symmetric in the high field limit . distributions at selected magnetic fields ( t=80 k).(b ) magnetic field dependence of the mean switching value , @xmath32 ( red squares ) , and the width of @xmath9 distributions , @xmath22 ( black circles).,width=321 ] ) ( a ) and the magnetization ( @xmath34 ) ( b ) on the out - of - plane magnetic field applied.,width=302 ] we consider whether this magnetic field dependence of the switching originates with some dependence of the bulk resistance or contact resistances , as this would alter the electric field distribution in the channel . since @xmath9 scales with the channel length , @xmath14 @xcite _ i.e. _ , as does the resistance of the channel ( @xmath35 ) , one might expect an _ increase _ in @xmath9 could result from an increase in the device resistance with applied magnetic field . however , fe@xmath36o@xmath37 has a negative magnetoresistance ( mr ) @xcite . figure [ r_m_vs_h ] a shows an example of normalized resistance dependence , @xmath38 , on the out - of - plane magnetic field at 80 k. the resistance remains effectively unchanged up to @xmath170.4 t and then decreases as @xmath39 increases . thus , when @xmath32 and @xmath22 experience the predominance of their changes upon @xmath30 application ( @xmath40 , see figure [ vsw_h_dep ] ) , the resistance of the device either stays constant or decreases . in the @xmath30 range when @xmath20 experiences significant changes ( see figure [ r_m_vs_h ] a ) , @xmath32 and @xmath22 remain essentially unchanged ( figure [ vsw_h_dep ] ) . therefore , the shift of @xmath32 in the presence of @xmath30 does not originate from the change in the resistance value of the device . another fe@xmath36o@xmath37 film parameter effected by @xmath30 is the magnetization of the film . figure [ r_m_vs_h ] shows the normalized out - of - plane magnetization , @xmath34 , as a function of the out of plane @xmath30 , where @xmath41 is the magnetization of the film and @xmath42 is the saturated magnetization . this data is consistent with prior measurements on magnetite films@xcite while we do not know the microscopic arrangement of @xmath43 in the film in the absence of an external @xmath44 , magnetostatic energy considerations mean that @xmath43 under that condition lies in the plane of the film . the @xmath30 range over which @xmath41 is fully reoriented out of the plane ( up to 1 t ) matches the @xmath30 range of changes in the position of @xmath32 and @xmath45 ( fig . [ vsw_h_dep ] b ) . this suggests ( though does not prove ) that the switching kinetics parameters @xmath32 and @xmath45 , and therefore the stability of the gapped , low temperature , insulating state is tied the magnetization direction of magnetite films . this observation is intriguing because it is not clear how the nonequilibrium breakdown of the low temperature state would be coupled to the magnetization . possible factors include magnetoelastic effects such as magnetostriction@xcite ( @xmath17 parts in @xmath46 per tesla ) affecting the tunneling matrix element between @xmath47-site iron atoms ; and spin - orbit coupling playing a similar role@xcite . there have been reports of significant magnetoelectric and multiferroic effects in magnetite @xcite , and a recent calculation @xcite argues that these originate through the interplay of orbital ordering and on - site spin - orbit interactions of the b - site electrons . in this picture , reorientation of the spin distorts the partially filled minority - spin orbitals occupied on the b - site ( formally ) fe@xmath3 ions . such a distortion would be a natural explanation for the observed correlation between @xmath43 and the kinetics of the electric field - driven breakdown of the ordered state , which directly involves the motions of those charge carriers . it is unclear how this kind of spin - orbit physics would explain the evolution of the @xmath9 distribution , however . it would also be worth considering whether there is any correlation between the characteristics of the switching distributions reported here , and the recently observed glassy relaxor ferroelectric relaxations in bulk magnetite crystals@xcite . additional , detailed experiments as a function of directionality of @xmath44 , @xmath43 , and crystallographic orientation should be able to test these alternatives . with the existing ( 100 ) films , studies of @xmath32 and @xmath45 as a function of @xmath30 in the plane as well as perpendicular to the plane should be able to access the tensorial form of the @xmath30 dependence . comparison with appropriately directed @xmath41 vs. @xmath30 data as a function of temperature would be a clear test of whether the observed agreement between @xmath30-field scales ( in @xmath9 and reorientation of @xmath43 is coincidental . further measurements on films grown with different crystallographic orientations would serve as a cross - check . it is important to note , however , that the acquisition of such data is very time intensive due to the need to acquire many thousands of switching events . in turn , there is a companion requirement of extremely good device stability , to avoid irreversible changes in the metal configuration over the thousands of switching cycles . we have studied the statistical distribution of the electric field needed for breakdown of the low temperature state of fe@xmath0o@xmath1 . the distribution of critical switching voltages moves to higher voltages and broadens , as @xmath10 is reduced . this broadening is consistent with phenomenology in other nonequilibrium experimental systems incorporating disorder and thermal runaway effects . the breakdown distributions are altered by modest external magnetic fields normal to the film , suggesting a need for further experiments to understand the connection between magnetization and breakdown of the correlated state . the authors acknowledge valuable conversations with paul goldbart and david pekker . this work was supported by the us department of energy grant de - fg02 - 06er46337 . dn also acknowledges the david and lucille packard foundation and the research corporation . rgss and ivs acknowledge the science foundation of ireland grant 06/in.1/i91 . mcqueeney r j , yethiraj m , montfrooij w , gardner j s , metcalf p and honig j m 2006 investigation of the presence of charge order in magnetite by measurement of the spin wave spectrum _ phys . rev . b _ * 73 * 174409 schlappa j , schssler - langeheine c , chang c f , ott h , tanaka a , hu z , haverkort m w , scheirle e , weschke e , kaindl g and tjeng l h 2008 direct observation of @xmath48 orbital ordering in magnetite _ phys . rev . lett . _ * 100 * , 026406 fursina a a , sofin r g s , shvets i v and natelson d 2010 interfacial transport properties between a strongly correlated transition metal oxide and a metal : contact resistance in fe@xmath0o@xmath1/m ( m = cu , au , pt ) nanostructures _ phys . rev . b _ * 82 * , 245112 koblischka - veneva a , koblischka m r , zhou y , murphy s , mucklich f , hartmann u and shvets i v 2007 electron backscatter diffraction analysis applied to [ 0 0 1 ] magnetite thin films grown on mgo substrates _ j. magn . mater . _ * 316 * , 663 - 665 arora s k , sofin r g s , shvets i v and luysberg m 2006 anomalous strain relaxation behavior of fe@xmath0o@xmath1/mgo ( 100 ) heteroepitaxial system grown using molecular beam epitaxy _ phys . _ * 100 * , 073908 varga r , garca k l , zhukov a , vzquez m and vojtanik p 2003 temperature dependence of the switching field and its distribution function in fe - based bistable microwires _ appl . lett . _ * 83 * , 2620 - 2622 sahu m , bae m h , rogachev a , pekker d , wei t c , shah n , goldbart p m and bezryadin a 2009 individual topological tunnelling events of a quantum field probed through their macroscopic consequences _ nature phys . _ * 5 * , 503 - 508 de teresa j m , fernndez - pacheco a , morellon l , orna j , pardo j a , serrate d , algarabel p a , ibarra m r 2007 magnetotransport properties of fe@xmath0o@xmath1 thin films for applications in spin electronics _ micro . engn . _ * 84 * , 1660 - 1664	the insulating state of magnetite ( fe@xmath0o@xmath1 ) can be disrupted by a sufficiently large dc electric field . pulsed measurements are used to examine the kinetics of this transition . histograms of the switching voltage show a transition width that broadens as temperature is decreased , consistent with trends seen in other systems involving `` unpinning '' in the presence of disorder . the switching distributions are also modified by an external magnetic field on a scale comparable to that required to reorient the magnetization .
nucleosynthesis via rapid neutron capture , the @xmath0-process , is a major mechanism for producing the elements heavier than fe @xcite . understanding this process requires knowledge of properties such as masses , @xmath11-decay lifetimes , and neutron - capture cross sections for a large number of extremely neutron - rich nuclei far from stability ( e.g. , @xcite ) . most of this nuclear input is beyond the reach of experiments in the foreseeable future and , therefore , must be calculated with guidance from existing data and from measurements to be made at rare - isotope beam facilities such as the gsi facility for antiproton and ion research ( fair ) , the facility for rare isotope beams ( frib ) , the heavy ion research facility in lanzhou cooling storage ring ( hirfl - csr ) , and rikagaku kenkyusho ( riken , institute of physical and chemical research , japan ) . in this paper we explore the importance of nuclear masses in estimating the temperature and neutron density conditions required for a specific @xmath0-process scenario , where neutron - capture reactions are in equilibrium with the reverse photodisintegration reactions ; i.e. , there is @xmath12 equilibrium ( e.g. , @xcite ) . using four nuclear mass models , we show that the required conditions can be determined mostly from the neutron - separation energies for a small number of critical nuclei with @xmath13 , 82 , and 126 closed neutron shells . for each model , we further determine the best - fit sets of conditions to reproduce the @xmath0-process abundance pattern ( @xmath0-pattern ) inferred for the solar system and observed in metal - poor stars of the milky way halo . this allows us to draw several interesting conclusions regarding the production of different parts of the overall @xmath0-pattern in an @xmath12 equilibrium scenario . we also illustrate the effects of nuclear mass uncertainties on the required @xmath0-process conditions and identify the key nuclei that have the largest impact and , therefore , are important candidates for precise mass measurements at rare - isotope beam facilities . we first give a brief overview of the @xmath0-process . detailed reviews can be found in ref . historically , the abundance distribution of nuclei in the solar system played a crucial role in studies on the origin of the elements @xcite . one of the prominent features of this distribution is the presence of three sets of double peaks in the region beyond the fe group nuclei . this was recognized as signatures of two distinct processes of neutron capture : a slow ( @xmath14 ) one encountering the @xmath13 , 82 , and 126 closed neutron shells in the stable region and a rapid ( @xmath0 ) one encountering the same in the extremely neutron - rich region of the nuclear chart @xcite . specifically , the peaks at mass numbers @xmath1 , 130 , and 195 were produced by the @xmath0-process and represent the crucial features of the solar @xmath0-pattern , which is derived by subtracting the @xmath14-process contributions from the net solar abundances ( e.g. , @xcite ) . in order to fully understand the @xmath0-process , we need conditions such as temperature and neutron density in the associated astrophysical environments in addition to the properties of a large number of extremely neutron - rich nuclei . neither the astrophysical nor the nuclear input is firmly established , although much progress has been made over the past two decades @xcite . proposed astrophysical sites for the @xmath0-process include neutrino - driven winds from proto - neutron stars formed in core - collapse supernovae ( ccsne ) @xcite , shocked surface layers of o - ne - mg cores associated with low - mass ccsne @xcite , winds from accretion disks of black holes formed in high - mass ccsne @xcite , he shells of metal - poor ccsne @xcite , and ejecta from neutron star mergers @xcite . there are large uncertainties in the conditions associated with all ccsne environments due to the substantial uncertainties in modeling such environments ( e.g. , @xcite ) , especially when neutrino transport in hot and dense nuclear matter is considered ( e.g. , @xcite ) . while recent studies lend much support to neutron star mergers being an @xmath0-process site @xcite , it remains to be seen whether such models are consistent with the history of @xmath0-process enrichments in the milky way and in its satellite dwarf galaxies ( e.g. , @xcite ) . further , how sensitive these models are to the uncertainties in the current understanding of the nuclear equation of state ( e.g. , @xcite ) and to the numerical treatment of the merger dynamics remains to be studied in detail . the conditions in the astrophysical environments relevant for the @xmath0-process ultimately boil down to the seed nuclei for neutron capture at the beginning of the process and the temperature @xmath15 and the neutron ( number ) density @xmath16 as functions of time @xmath17 during the process ( e.g. , @xcite ) . in cases where neutrino interactions are important , the time evolution of neutrino fluxes and energy spectra is also required ( e.g. , @xcite ) . in the rest of the paper we will ignore neutrinos and focus on a broad class of astrophysical environments where matter undergoes @xmath0-processing at @xmath18 k and @xmath19 @xmath20 . for such high temperatures and neutron densities , previous studies have shown that @xmath12 equilibrium is achieved ( e.g. , @xcite ) . in this equilibrium , the abundance distribution in each isotopic chain at a specific proton number @xmath21 is almost always strongly peaked at one nucleus . this is referred to as a waiting - point ( wp ) nucleus because , upon reaching it , the @xmath0-process must wait for it to @xmath11-decay before producing heavier nuclei . under this so - called wp approximation , the @xmath0-process path is defined by all the wp nuclei heavier than the seed nuclei , and the progress along this path is regulated by the @xmath11-decay of these wp nuclei . so long as this approximation is valid , there is no need to follow neutron capture and photodisintegration reactions , which greatly simplifies the @xmath0-process calculation . due to the equilibrium between neutron capture and photodisintegration reactions , the abundance ratio between two neighboring isotopes is given by the saha equation ( e.g. , @xcite ) : @xmath22,\ ] ] where @xmath23 is the planck constant , @xmath24 is the atomic mass unit , @xmath25 is the boltzmann constant , @xmath26 indicates a nucleus with proton number @xmath21 and mass number @xmath27 , and @xmath28 , @xmath29 , and @xmath30 denote the number abundance , partition function , and neutron separation energy of the appropriate nucleus , respectively . for a specific isotopic chain , the corresponding wp nucleus has the largest abundance and is determined by the partition functions and neutron separation energies of the relevant nuclei for fixed @xmath31 and @xmath32 . as can be seen from the exponential dependence on the neutron separation energy in eq . ( [ equilibrium ] ) , nuclear masses are among the most important input for modeling the @xmath0-process . the other crucial input is @xmath11-decay lifetimes of the relevant nuclei . over the past two decades , tremendous progress has been made in measuring nuclear properties relevant for the @xmath0-process . for example , the @xmath11-decay half - lives of 38 very neutron - rich isotopes bordering the @xmath0-process path have been measured recently @xcite . in addition , the masses of a group of nuclei including @xmath33zn @xcite and @xmath34cd @xcite have been measured with a very high accuracy @xcite . meanwhile , considerable advance has been made in the theoretical investigation of nuclear masses . the four nuclear mass models used in this paper span from the macroscopic - microscopic kind , represented by the finite - range droplet model ( frdm ) @xcite and a more recent weizscker - skyrme ( ws * ) model @xcite , to the microscopic kind , represented by the skyrme - hartree - fock - bogolyubov mean - field ( hfb-17 ) model @xcite and the relativistic mean - field ( rmf ) model @xcite . these models can reproduce the experimentally known neutron separation energies with a root - mean - square ( rms ) deviation of 0.399 ( frdm ) , 0.332 ( ws * ) , 0.506 ( hfb-17 ) , and 0.653 ( rmf ) mev , respectively . based on the above overview , there are two frontiers of @xmath0-process research : one focusing on the search for the astrophysical sites and quantification of the conditions therein and the other on acquiring a reliable database for the relevant nuclear input . observations of elemental abundances in metal - poor stars of the milky way halo ( see @xcite for a review ) have shed important light on the @xmath0-process sites ( e.g. , @xcite ) . the observed @xmath0-patterns also provide an important test of the basic soundness of the nuclear input ( e.g. , @xcite ) . of course , the astrophysical and the nuclear input must be coupled together in order to produce an @xmath0-pattern for comparison with observations . with substantial uncertainties in the current understanding of both the @xmath0-process sites and the nuclear input , parametrization of the astrophysical conditions is often used in exploring the effects of nuclear input on @xmath0-process production ( e.g. , @xcite ) . as a practical matter , we adopt the classical approach of using @xmath31 , @xmath32 , and the corresponding neutron irradiation time @xmath35 along with the wp approximation ( e.g. , @xcite ) to carry out our @xmath0-process calculations below . our main purpose is to explore the effects of the four nuclear mass models mentioned above on the @xmath31 and @xmath32 conditions required for @xmath0-process nucleosynthesis . a number of other studies on how the nuclear input impacts @xmath0-process nucleosynthesis have been carried out recently . the influence of nuclear properties from different mass models on the final @xmath0-pattern was analyzed in ref . the effect of neutron capture rates for nuclei near the @xmath4 peak on the overall @xmath0-pattern was investigated in ref . the sensitivity of the calculated @xmath0-pattern to the combined effects of the long - term dynamic evolution of the astrophysical environment and the nuclear input was explored in ref . the effect of long - range correlations for nuclear masses on the production of nuclei immediately below and in the @xmath36 peak was studied in ref . @xcite . in all of the above studies parametric astrophysical models that allow @xmath31 and @xmath32 to evolve with time were adopted . these models are more realistic but the choice of parameters is not so straightforward . in the future , we plan to use similar models to study the interplay between the astrophysical and the nuclear input during the @xmath0-process . our goal here is to explore the @xmath31 and @xmath32 conditions required for the @xmath0-process in an @xmath12 equilibrium scenario and the effects of nuclear mass models on these conditions . for this limited goal , we feel that the classical approach based on the wp approximation is adequate . we give a detailed discussion of the wp approximation in sec . [ sec : model ] . using this approximation along with four nuclear mass models , we derive for each model the @xmath31 and @xmath32 conditions that are required for producing the abundance peaks at @xmath1 , 130 , and 195 as observed in the solar system . in sec . [ sec : patterns ] we describe the classical approach to simulate the @xmath0-process and use this approach to determine the sets of conditions that can best reproduce the solar @xmath0-pattern and the @xmath0-patterns observed in metal - poor stars for each of the adopted nuclear mass models . we discuss our results and give conclusions in sec . [ sec : discuss ] . while there are substantial uncertainties in both the @xmath0-process sites and the relevant nuclear input , an essential feature of the @xmath0-process is considered robust : the observed abundance peaks at @xmath1 , 130 , and 195 correspond to the intrinsic properties of extremely neutron - rich nuclei with @xmath13 , 82 , and 126 closed neutron shells that are produced in the @xmath0-process . as discussed in sec . [ sec : introduction ] , when @xmath12 equilibrium is achieved , the total abundance of an isotopic chain is concentrated in the corresponding wp nucleus . the @xmath11-decay lifetimes of the wp nuclei then regulate the abundance pattern resulting from an @xmath0-process episode . in particular , the much longer @xmath11-decay lifetimes of extremely neutron - rich nuclei with closed neutron shells than those without produce peaks in @xmath0-patterns . consequently , in order to produce the observed peaks in @xmath0-patterns under the wp approximation , nuclei with @xmath13 , 82 , and 126 closed neutron shells and with @xmath1 , 130 , and 195 , respectively , must be among the wp nuclei . the critical wp ( cwp ) nuclei were discussed by earlier studies ( e.g. , @xcite ) . for the present work , we select the cwp nuclei listed in table [ tabcwp ] based on similar considerations to these studies . below we follow the spirit of previous studies to derive the conditions required for @xmath0-process nucleosynthesis under the wp approximation by considering the properties of the cwp nuclei . .critical waiting - point nuclei . [ cols="<,<",options="header " , ] to be quantitative , we define a wp nucleus @xmath37 as one that has an abundance @xmath38 where the sum over @xmath27 gives the total abundance of the corresponding isotopic chain . for specific @xmath31 and @xmath32 , we can use the above criterion and the relative abundance @xmath39 given by eq . ( [ equilibrium ] ) to determine @xmath37 from the nuclear partition functions and neutron separation energies provided by a model . conversely , we can also determine the @xmath31 and @xmath32 conditions required by a specific wp nucleus . as can be seen from eq . ( [ equilibrium ] ) , the predominant dependence of @xmath39 is on the neutron separation energy , which can be calculated from a nuclear mass model . we ignore the small differences in the nuclear partition function and in the mass number and rewrite eq . ( [ equilibrium ] ) as @xmath40,\ ] ] where @xmath41 = t_9\left[2.79+\frac{1.5\log_{10 } t_9-\log_{10}(n_n/10^{20}\ { \rm cm}^{-3})}{5.04}\right]\,\rm{mev}.\ ] ] in the second equality of eq . ( [ sn0 ] ) , @xmath42 is @xmath31 in units of @xmath43 k. equation ( [ equilibrium2 ] ) is used in the calculations below . as an example , we adopt the ws * mass model to calculate the sets of @xmath42 and @xmath32 within the ranges @xmath44 and @xmath45 @xmath20 that are required by the @xmath3 cwp nuclei . the results are shown in fig . [ figwspure82 ] . for a specific @xmath42 , the values of @xmath32 between two identical symbols in this figure would allow the corresponding nucleus to have @xmath46 of the total abundance of its isotopic chain . in order to accommodate all the @xmath3 cwp nuclei , the common range of @xmath32 for a specific @xmath42 is bounded from below by @xmath47ru ( filled circle ) and from above by @xmath34cd ( filled triangle ) . this range of @xmath32 changes with @xmath42 and is shown as the shaded band in fig . [ figwspure82 ] . this band represents the @xmath42-@xmath32 conditions required by the @xmath3 cwp nuclei . -@xmath32 conditions required by the @xmath5 cwp nuclei based on the ws * mass model . for a specific @xmath42 , the values of @xmath32 between two identical symbols would allow the corresponding nucleus to have @xmath46 of the total abundance of its isotopic chain . the conditions indicated by the shaded band are required to accommodate all the @xmath5 cwp nuclei . ] while the neutron separation energies of a large number of nuclei from the ws * mass model are used in the above calculations , the results in fig . [ figwspure82 ] are determined effectively by the two - neutron separation energies of the @xmath3 cwp nuclei and their @xmath48 isotopes due to nuclear systematics . this can be understood as follows . because of the effect of pairing on neutron binding , all wp nuclei have even @xmath49 . the relative abundance of two neighboring even-@xmath49 isotopes can be obtained from eq . ( [ equilibrium2 ] ) as @xmath50,\ ] ] where @xmath51 denotes the two - neutron separation energy . the values of @xmath52 for isotopes of ru , rh , pd , ag , and cd ( @xmath53 ) are shown as functions of @xmath49 in fig . [ figs2n ] , which exhibits the general trend that @xmath51 essentially monotonically decreases with @xmath49 for a specific isotopic chain . based on this aspect of nuclear systematics , it can be seen from eq . ( [ equibibriumeven ] ) that the abundance of an even-@xmath49 isotope increases with @xmath49 [ @xmath54 until @xmath55 falls below @xmath56 , from which point on it decreases with @xmath49 [ @xmath57 . therefore , the abundance of even-@xmath49 isotopes peaks at the nucleus @xmath37 , for which @xmath58 the above equation effectively defines a wp nucleus ( e.g. , @xcite ) and can be used to determine the @xmath42-@xmath32 conditions required by a specific wp nucleus . for example , @xmath34cd has @xmath59 mev while @xmath10cd has @xmath60 mev @xcite . so the @xmath42-@xmath32 conditions corresponding to @xmath61 mev [ the band between dashed lines labeled as @xmath62 in fig . [ figs2n ] ] are required for @xmath34cd to be a wp nucleus . likewise , the shaded band labeled as @xmath63 in fig . [ figs2n ] corresponds to the conditions required to accommodate all the @xmath3 cwp nuclei ( shaded band in fig . [ figwspure82 ] ) . for even-@xmath49 isotopes of ru , rh , pd , ag , and cd ( @xmath64 ) in the ws * model shown in terms of @xmath52 as a function of @xmath49 around @xmath5 . the band between the dashed lines labeled as @xmath65 corresponds to the @xmath42-@xmath32 conditions required for @xmath34cd to be a wp nucleus . likewise , the shaded band labeled as @xmath66 corresponds to the conditions required to accommodate all the @xmath3 cwp nuclei ( shaded band in fig . [ figwspure82 ] ) . see text for details . ] the calculations in sec . [ sec : nsep ] can be generalized to determine the @xmath42-@xmath32 conditions required by the @xmath13 , 82 , and 126 cwp nuclei , respectively , for any specific nuclear mass model . the results for the @xmath13 cwp nuclei are presented for the frdm , ws * , and rmf models in fig . [ figmodel50](a ) , which clearly show that the required conditions change with models . similar to the case of the @xmath3 cwp nuclei ( figs . [ figwspure82 ] and [ figs2n ] ) discussed in sec . [ sec : nsep ] , the upper curve for each model in fig . [ figmodel50](a ) is effectively determined by the two - neutron separation energy of the lightest @xmath13 cwp nucleus @xmath7ni and the lower curve by that of the @xmath67 isotope @xmath8zn of the heaviest @xmath13 cwp nucleus @xmath33zn . therefore , the large differences among the conditions required by the @xmath13 cwp nuclei for different models can be traced to the differences in the two - neutron separation energies of @xmath7ni and @xmath8zn provided by these models . in particular , the differences for @xmath8zn among the models appear to be substantially larger than those for @xmath7ni . we also note that no conditions can be found to accommodate all the @xmath13 cwp nuclei for the hfb-17 model , for which the odd - even effects in the neutron separation energy for ni , cu , and zn isotopes around @xmath68 are larger by @xmath691.5 mev than those for the frdm , ws * , and rmf models . -@xmath32 conditions required by the @xmath13 cwp nuclei . ( a ) the band between two curves of the same kind represents the required conditions based on the corresponding mass model . note the large differences among the results for the three indicated models . note also that no conditions can be found to accommodate all the @xmath13 cwp nuclei based on the hfb-17 model . ( b ) same as ( a ) , but for those nuclei whose masses are tabulated in the latest atomic mass evaluation ame2011-preview @xcite , the model predictions are replaced by the tabulated values . all four models now give the same conditions required by the @xmath13 cwp nuclei . ] noting that the two - neutron separation energies of @xmath7ni and @xmath8zn can be calculated from the masses tabulated in the latest atomic mass evaluation ame2011-preview @xcite , we augment the nuclear mass models by using the tabulated values in ame2011-preview when they are available to replace the corresponding model predictions . remarkably , all four models , including the hfb-17 model , now give the same conditions required by the @xmath13 cwp nuclei as shown in fig . [ figmodel50](b ) . we find that the changes between figs . [ figmodel50](a ) and [ figmodel50](b ) are caused dominantly by the use of the tabulated masses of @xmath6ni to @xmath7ni and @xmath7zn to @xmath8zn , which confirms the crucial roles of the two - neutron separation energies of @xmath7ni and @xmath8zn in determining the conditions required by the @xmath13 cwp nuclei . in the calculations below , we use the frdm , ws * , hbf-17 , and rmf models that are augmented by ame2011-preview . we calculate the conditions required by the @xmath3 and 126 cwp nuclei , respectively , as in the case of the @xmath13 cwp nuclei . the results are summarized in fig . [ figmodelall ] . it can be seen that the conditions required by the @xmath3 cwp nuclei ( shaded band ) are essentially converged for the four augmented nuclear mass models just like those required by the @xmath13 cwp nuclei ( horizontally hatched band between solid curves ) . in contrast , the conditions required by the @xmath70 cwp nuclei ( vertically hatched band between dashed curves ) are still strongly dependent on models . this is because these nuclei and the majority of those in the nearby region of the nuclear chart are still out of the reach of experiments while theoretical predictions for their masses involve dramatic extrapolations with large uncertainties ( e.g. , @xcite ) . -@xmath32 conditions required by the @xmath68 ( horizontally hatched band between solid curves ) , 82 ( shaded band ) , and 126 ( vertically hatched band between dashed curves ) cwp nuclei , respectively , for four nuclear mass models : ( a ) rmf , ( b ) hfb-17 , ( c ) frdm , and ( d ) ws. see text for details . ] figure [ figmodelall ] resembles a phase diagram in terms of three bands for the production of the peaks at @xmath1 , 130 , and 195 in @xmath0-patterns that correspond to the @xmath13 , 82 , and 126 cwp nuclei . for the @xmath42-@xmath32 conditions inside the nonoverlap region of a band , only a single peak can be produced . for those conditions inside the overlap region of two bands , it is possible to produce two peaks simultaneously . for the rmf model only , there is a very thin sliver where three bands overlap . consequently , we consider it very unlikely that three peaks can be produced simultaneously . a close examination of fig . [ figmodelall ] shows that the @xmath42-@xmath32 conditions required by the @xmath68 cwp nuclei ( horizontally hatched band between solid curves ) are distinct from those required by the @xmath70 cwp nuclei ( vertically hatched band between dashed curves ) for the frdm , hfb-17 , and ws models . these two sets of conditions overlap only slightly for the rmf model . this suggests that the peaks at @xmath1 and 195 in @xmath0-patterns are not produced simultaneously . their production may differ in the time of occurrence within the same astrophysical site or in the astrophysical site itself . in contrast , there is large overlap between the conditions required by the @xmath68 and 82 cwp nuclei ( shaded band ) for the four models considered . in addition , there is slight to significant overlap between the conditions required by the @xmath3 and 126 cwp nuclei for all the models . therefore , it is possible to produce the peaks at @xmath1 and 130 or those at @xmath4 and 195 simultaneously . the above results will be examined by detailed @xmath0-process calculations in sec . [ sec : patterns ] . our main goal here is to explore the effects of nuclear masses on the conditions required for @xmath0-process nucleosynthesis under the wp approximation . as discussed in sec . [ sec : model ] , these conditions are mostly set by the neutron separation energies , which we calculate from four nuclear mass models augmented by the latest atomic mass evaluation ame2011-preview . as confirmation of these results , we calculate the @xmath0-patterns produced under the conditions shown in fig . [ figmodelall ] and compare them with those inferred for the solar system and observed in metal - poor stars . as the range of conditions shown in fig . [ figmodelall ] is rather broad , we sample these conditions at a fixed temperature . specifically , we use @xmath71 and @xmath72@xmath73 , @xmath74@xmath75 , and @xmath76@xmath77 @xmath20 as typical conditions required by the @xmath13 , 82 , and 126 cwp nuclei , respectively . as noted in sec . [ sec : cond ] , there is overlap between these sets of conditions , which can lead to coproduction of two peaks in the @xmath0-pattern . we carry out an @xmath0-process calculation using the classical approach ( e.g. , @xcite ) as follows . we take the seed nucleus to be @xmath78fe ( @xmath79 ) . we assume that some material with an initial abundance @xmath80 of @xmath78fe is irradiated with neutrons at fixed @xmath31 and @xmath32 for a time @xmath35 . we then use eq . ( [ equilibrium2 ] ) along with a nuclear mass model to calculate the fractional abundance @xmath81 for all the nuclei in each of the isotopic chains with @xmath82 , where @xmath83 note that for fixed @xmath31 and @xmath32 , @xmath81 is also fixed . using the fractional abundances , we calculate the effective @xmath11-decay rate of an isotopic chain as @xmath84 where @xmath85 is the @xmath11-decay rate of the nucleus @xmath26 . for all our calculations , we use @xmath85 from the experimental data in ref . @xcite for the nuclei with measurements and from the theoretical estimates in ref . @xcite based on the frdm+qrpa method for those without . we then solve the set of equations @xmath86 where the dot denotes the derivative with respect to time @xmath17 , and @xmath87 is the total abundance of the isotopic chain with proton number @xmath21 at time @xmath17 . the initial conditions are @xmath88 for @xmath79 and 0 for @xmath89 . we assume that the @xmath0-process freezes out instantaneously at @xmath90 . the freeze - out abundance of the nucleus @xmath26 is @xmath91 the final abundance distribution from an @xmath0-process episode is obtained by following the @xmath11 and @xmath92 decays of all the nuclei in the freeze - out distribution . we include @xmath11-delayed emission of up to three neutrons @xcite , which has the important effect of smoothing the final @xmath0-pattern . the data on @xmath92-decays are taken from the national nuclear data center @xcite . fission is ignored in all the calculations . note that the wp approximation is implicit in the classical approach as @xmath87 is dominated by the corresponding wp nucleus with @xmath93 . for comparison with the @xmath0-patterns inferred for the solar system and observed in metal - poor stars , we need to superpose the results from many @xmath0-process episodes described above . as we take @xmath71 for all the calculations , we denote each episode by its @xmath32 . the neutron irradiation time @xmath94 and the weight @xmath95 for each episode are taken to be @xmath96 where @xmath97 , @xmath98 , @xmath99 , and @xmath100 are parameters to be determined by a least - squares fit to the @xmath0-pattern used for comparison . while such a superposition procedure is a crude approximation to estimate @xmath0-patterns produced by astrophysical sources , it can still provide some useful information on the conditions that the actual @xmath0-process sites must fulfill @xcite . for this reason , this procedure has been used extensively in @xmath0-process studies ( e.g. , @xcite ) . we first use the classical approach to reproduce the solar @xmath0-pattern for @xmath101 @xcite ( see also , e.g. , @xcite ) , which is shown as the dashed curve in fig . [ figfitsolar1 ] . we consider a superposition of nine neutron densities ( equidistant on a @xmath102 scale ) within the range @xmath103 @xmath20 , which corresponds to the typical conditions required by the @xmath70 cwp nuclei for @xmath71 ( see fig . [ figmodelall ] ) . the best - fit results ( hereafter `` fit i '' ) for the four adopted nuclear mass models are shown as the solid curves in fig . [ figfitsolar1 ] . note that although the fits are performed for the solar isotopic @xmath0-pattern , the patterns shown in fig . [ figfitsolar1 ] are for the corresponding elemental abundances . it can be seen that the solar @xmath0-pattern from the peak at @xmath4 ( @xmath104 , te ) to that at @xmath36 ( @xmath105 , pt ) is reproduced rather well for the frdm , hbf-17 , and ws * models . for the rmf model , the rare - earth elements with @xmath10670 ( dy , ho , er , tm , and yb ) are severely underproduced . this deficiency may reflect the necessity to adopt improved rmf parameter sets ( e.g. , pc - pk1 @xcite , which provides a much better description for the properties of nuclear ground and excited states @xcite ) or that the classical approach is inadequate to give a full description of @xmath0-process nucleosynthesis ( e.g. , instantaneous freeze - out is not a good approximation @xcite ) . in any case , the peaks at @xmath4 and 195 are reproduced adequately for all four mass models , which suggests that the typical conditions required by the @xmath70 cwp nuclei can indeed produce both these peaks . as discussed in sec . [ sec : cond ] , this is because there is significant overlap between the conditions required by the @xmath3 and 126 cwp nuclei ( see fig . [ figmodelall ] ) . -process approach using four nuclear mass models : ( a ) rmf , ( b ) hfb-17 , ( c ) frdm , and ( d ) ws. the filled circles represent the average data on @xmath0-ii stars of the milky way halo and the dashed curve is the solar @xmath0-pattern translated to pass through the eu data . the filled square gives the te abundance recently measured in the metal - poor star bd + 17@xmath1073248 ( shifted according to the observed te / eu ratio ) . the calculated solid curves ( fit i ) are the best - fit results to reproduce the solar isotopic @xmath0-pattern for @xmath101 with @xmath71 and @xmath103 @xmath20 , which are the typical conditions required by the @xmath70 cwp nuclei . see text for details . ] observations show that ba ( @xmath108 ) and heavier elements in many metal - poor stars of the milky way halo follow the solar @xmath0-pattern rather closely @xcite . the values of @xmath109 , where @xmath110 and @xmath111 represent the abundance of element e and hydrogen respectively , obtained for the elements with @xmath112 by averaging the data @xcite on two such `` @xmath0-ii '' stars , cs 22892052 and cs 31082001 , are shown as filled circles in fig . [ figfitsolar1 ] . the data on pb ( @xmath113 ) and th ( @xmath114 ) for cs 22892052 are also shown . the solar @xmath0-pattern has been translated to pass through the data on eu ( @xmath115 ) and can be seen to represent the pattern for @xmath116 in @xmath0-ii stars very well . recently , te ( @xmath117 ) has been measured in a group of metal - poor stars ( bd + 17@xmath1073248 , hd 108317 , and hd 128279 @xcite and hd 160617 @xcite ) . this extends the comparison of @xmath0-patterns in metal - poor stars with the solar @xmath0-pattern to include an element in the peak at @xmath118 . the data on te for bd + 17@xmath1073248 ( shifted according to the observed te / eu ratio ) is shown as the filled square in fig . [ figfitsolar1 ] . it can be seen that the te data is consistent with the solar @xmath0-pattern and with the coproduction of the peaks at @xmath4 and 195 under the typical conditions required by the @xmath70 cwp nuclei . the above results are also in agreement with previous studies ( e.g. , @xcite ) , in which it was concluded that the @xmath0-process responsible for the elements with @xmath119 is characterized by neutron densities of @xmath120@xmath121 @xmath20 . the fit i results shown in fig . [ figfitsolar1 ] can not adequately reproduce the abundances of the elements with @xmath122 in @xmath0-ii stars ( especially when the rmf model is used ) . further , the elements in the peak at @xmath1 ( @xmath123 ) of the solar @xmath0-pattern are severely underproduced by these calculations . additional @xmath0-process contributions to or alternative sources for the elements below the peak at @xmath4 are thus required and this issue has been under active investigation @xcite . here we explore the possibility that there are additional contributions from @xmath0-process nucleosynthesis under the conditions required by the @xmath13 cwp nuclei . we consider a superposition of six neutron densities ( equidistant on a @xmath102 scale ) within the range @xmath124 @xmath20 to best reproduce the solar isotopic @xmath0-pattern for @xmath125 ( hereafter `` fit ii '' ) . the results are shown as the solid curves in fig . [ figfitsolar2 ] . it can be seen that the conditions required by the @xmath13 cwp nuclei indeed can produce the peak at @xmath1 . however , it is also clear that the @xmath0-patterns from the peak at @xmath4 to that at @xmath36 inferred for the solar system and observed in @xmath0-ii stars require very different conditions from those for producing the peak at @xmath1 ( see sec . iv ) . , but the solid curves ( fit ii ) are the best - fit results to reproduce the solar isotopic @xmath0-pattern for @xmath125 with @xmath71 and @xmath124 @xmath20 , which are the typical conditions required by the @xmath13 cwp nuclei . see text for details . ] to find the best match to the @xmath0-pattern in @xmath0-ii stars , we consider a superposition of neutron densities in the two ranges adopted for fits i and ii . the results are shown as the solid curves in fig . [ figsumr - ii ] . it can be seen that fair agreement between the calculated and observed patterns is obtained for the frdm , hfb-17 , and ws models . note also that the calculated te abundances ( crosses ) for all four models are consistent with the newly measured value for the metal - poor star bd + 17@xmath1073248 . however , the trough at @xmath10670 is clearly problematic for the rmf model . this may be caused by nuclear shape transition before the @xmath70 closed neutron shell and the location of the transition region could have been assigned incorrectly in the rmf model @xcite . discrepancies can also be seen for ru , rh , and ag ( @xmath126 , 45 , and 47 , respectively ) for all four models . this issue needs to be addressed by detailed considerations of the astrophysical environments for the @xmath0-process and alternative sources for the elements below the peak at @xmath4 @xcite . -pattern in @xmath0-ii stars ( filled circles ) using a superposition of neutron densities within the ranges @xmath124 @xmath20 and @xmath103 @xmath20 for four nuclear mass models : ( a ) rmf , ( b ) hfb-17 , ( c ) frdm , and ( d ) ws. the crosses give the calculated te abundances , which are consistent with the measured value ( filled square ) for the metal - poor star bd + 17@xmath1073248 ( shifted according to the observed te / eu ratio ) . see text for details . ] in contrast to the @xmath0-ii stars , some metal - poor stars exhibit an @xmath0-pattern that is clearly different from the solar one . prominent examples are the metal - poor stars hd 122563 @xcite and hd 88609 @xcite , which have almost the same abundances for cu ( @xmath127 ) and heavier elements . the data on sr ( @xmath128 ) and heavier elements for hd 122563 are shown as the filled circles in fig . [ figfithd ] . relative to the solar @xmath0-pattern translated to pass through the eu data ( dashed curve ) , the elements below the peak at @xmath4 in this star are grossly overabundant . it was argued that in addition to an @xmath0-process source for producing a solar - like @xmath0-pattern from the peak at @xmath4 to that at @xmath36 , a very different source is required to explain the data for stars like hd 122563 @xcite . -pattern ( filled circles ) in the metal - poor star hd 122563 for four nuclear mass models : ( a ) rmf , ( b ) hfb-17 , ( c ) frdm , and ( d ) ws. these results use a superposition of neutron densities within the range @xmath129 @xmath20 , which corresponds to the typical conditions required by the @xmath3 cwp nuclei for @xmath71 . the dashed curve gives the solar @xmath0-pattern translated to pass through the filled circle for eu ( @xmath115 ) . see text for details . ] here we attempt to interpret the abundance pattern observed in hd 122563 using the classical @xmath0-process approach . we find that the conditions required by the @xmath5 cwp nuclei can best reproduce this pattern while those required by the @xmath13 and 126 cwp nuclei can not . the best - fit results use a superposition of ten neutron densities ( equidistant on a @xmath102 scale ) within the range @xmath129 @xmath20 , which corresponds to the typical conditions required by the @xmath3 cwp nuclei for @xmath71 ( see fig . [ figmodelall ] ) . these results are shown as the solid curves in fig . [ figfithd ] . it can be seen that an approximate overall match of the calculated with the observed abundances is obtained for all four nuclear mass models . therefore , we suggest that it is plausible to account for the abundance pattern in stars like hd 122563 by an @xmath0-process operating under the conditions required by the @xmath5 cwp nuclei . we note that fig . [ figfithd ] shows a clear difference in the calculated relative production of te and xe ( @xmath117 and 54 , respectively ) between the ws * and the other three mass models : these two elements are produced in approximately equal amount for the ws * model but xe is produced much more than te for the other three models . measurements of these two elements in hd 122563 would be extremely valuable in constraining nuclear mass models although they also represent a difficult challenge to spectroscopic observations . we have explored the effects of four nuclear mass models ( frdm , ws * , hbf-17 , and rmf ) on the conditions required by @xmath0-process nucleosynthesis under the wp approximation . as discussed in sec . [ sec : model ] , the required @xmath42-@xmath32 conditions are mostly determined by the two - neutron separation energies of the cwp nuclei with @xmath13 , 82 , and 126 and of those nuclei around them . figure [ figmodel50 ] shows the dramatic effect of using the tabulated values in the latest atomic mass evaluation ame2011-preview when they are available to replace the masses predicted by models . as noted in sec . [ sec : cond ] , the tabulated masses of @xmath6ni to @xmath7ni and @xmath7zn to @xmath8zn play crucial roles in determining the conditions required by the @xmath13 cwp nuclei . however , the tabulated masses of @xmath6ni to @xmath7ni and @xmath8zn are extrapolated rather than measured . to emphasize the effects of these masses on the conditions required by the @xmath13 cwp nuclei , we first repeat the calculations of sec . [ sec : model ] by varying the neutron separation energy of @xmath7ni within the estimated uncertainty of 0.946 mev @xcite while keeping the other input the same as for fig . [ figmodel50](b ) . the results are shown in fig . [ figerr](a ) . in comparison with fig . [ figmodel50](b ) , the lower bound on the region of the required @xmath42-@xmath32 conditions stays the same because this is determined by the two - neutron separation energy of @xmath8zn ( see sec . [ sec : cond ] ) , which is not changed . increasing the neutron separation energy of @xmath7ni by 0.946 mev raises the upper bound from the solid curve [ upper bound in fig . [ figmodel50](b ) ] to the dashed curve and decreasing this quantity by the same amount lowers it to the dotted curve . we then repeat the same calculations but vary the neutron separation energy of @xmath8zn within the estimated uncertainty of 0.401 mev @xcite . the effects on the lower bound on the region of the required @xmath42-@xmath32 conditions are shown in fig . [ figerr](b ) . note that if the neutron separation energy of @xmath7ni were lower than its tabulated value by 0.946 mev while that of @xmath8zn were higher by 0.401 mev , then it would be almost impossible to find any @xmath42-@xmath32 conditions to accommodate all the @xmath13 cwp nuclei . in any case , the significant effects of uncertainties in neutron separation energies on the required @xmath42-@xmath32 conditions shown in panels ( a ) and ( b ) of fig . [ figerr ] clearly demonstrate the importance of precise mass measurements for @xmath6ni to @xmath7ni and @xmath8zn . for ( a ) @xmath7ni , ( b ) @xmath8zn , ( c ) @xmath130 tb , and ( d ) @xmath131tm on the required @xmath42-@xmath32 conditions . the solid curves in panels ( a ) and ( b ) are the same as those in fig . [ figmodel50](b ) . the solid curves in panels ( c ) and ( d ) are the same as the dashed curves ( for the ws * model ) in fig . [ figmodelall](d ) . the shaded regions in each panel show the effects on the required @xmath42-@xmath32 conditions when the corresponding @xmath30 values are varied within the estimated uncertainties . see text for details . ] as in the case of @xmath13 cwp nuclei , we have also made a careful scan of the nuclear chart around the @xmath5 cwp nuclei and explored the effects of those nuclei with experimentally unknown or poorly measured masses on the @xmath42-@xmath32 conditions required by the @xmath5 cwp nuclei . when nuclear masses are not known experimentally , we have used the extrapolated masses and uncertainties as these have been proven to have a better predictive power than all available models @xcite . we have identified @xmath9cd and @xmath10cd as additional key nuclei for precise mass measurements . as can be seen from fig . [ figmodelall ] , the @xmath42-@xmath32 conditions required by the @xmath70 cwp nuclei depend strongly on the nuclear mass model . to assess the impact of uncertainties in nuclear mass models , we varied the neutron separation energies @xmath30 for the relevant nuclei within the known errors or the rms deviations of model predictions for the known masses . using the ws * model as an example , we show the effects of uncertainties in @xmath30 for @xmath130 tb and @xmath131tm on the required @xmath42-@xmath32 conditions in panels ( c ) and ( d ) of fig . [ figerr ] , respectively . it can be seen that the upper bound on these conditions changes very little when @xmath132 is varied within 0.332 mev ( 1 rms deviation for the ws * model ) but the lower bound is more sensitive to the same variation of @xmath133 . however , the conditions required by the @xmath13 [ figs . [ figerr](a ) and 9(b ) ] and 126 [ figs . [ figerr](c ) and 9(d ) ] cwp nuclei do not appear to overlap for the ws * model even when uncertainties in nuclear masses are considered . this suggests that , at least for this model , the @xmath1 and 195 peaks in the @xmath0-pattern are most likely produced under very different conditions . precise mass measurements and better calibrated mass models are needed to make this result more robust . in conclusion , we have estimated the temperature and neutron density conditions required for @xmath0-process nucleosynthesis under the wp approximation using four nuclear mass models augmented by the latest atomic mass evaluation ame2011-preview . we have shown that these conditions are mostly determined by the two - neutron separation energies of the @xmath13 , 82 , and 126 cwp nuclei and those around them . we have also identified some key nuclei including @xmath6ni to @xmath7ni , @xmath8zn , @xmath9cd , and @xmath10cd for precise mass measurements at rare - isotope beam facilities . based on the typical conditions required by the @xmath13 , 82 , and 126 cwp nuclei shown in fig . [ figmodelall ] , we have performed @xmath0-process calculations in the classical approach to reproduce the @xmath0-pattern inferred for the solar system and those observed in metal - poor stars of the milky way halo . we have found that ( 1 ) at least for the nuclear mass models considered here , the conditions required to produce the peak at @xmath1 differ greatly from those required to produce the solar @xmath0-pattern from the peak at @xmath4 to that at @xmath36 , which reflects that the @xmath42-@xmath32 conditions required by the @xmath13 and 126 cwp nuclei are very different ( especially for the ws * model ) ; ( 2 ) the solar @xmath0-pattern from the peak at @xmath4 to that at @xmath36 , which also closely describes the @xmath0-patterns in many metal - poor stars , can be reproduced under the conditions required by the @xmath70 cwp nuclei , which has significant overlap with those required by the @xmath3 cwp nuclei , thereby enabling coproduction of the peaks at @xmath4 and 195 ; ( 3 ) it is plausible to explain the overall @xmath0-patterns in metal - poor @xmath0-ii stars with a superposition of two sets of @xmath0-process conditions required by the @xmath68 and 126 cwp nuclei , respectively ; and ( 4 ) the non - solar - like @xmath0-pattern observed in metal - poor stars like hd 122563 can be accounted for by the @xmath0-process conditions required by the @xmath5 cwp nuclei . we note that similar results were also obtained by other earlier studies ( e.g. , @xcite ) . we recognize that the classical @xmath0-process approach leaves out many important details , such as the time evolution of temperature and neutron density , the finite duration of the freeze - out , and the breakdown of @xmath12 equilibrium during the freeze - out . we note that the impact of the details of the freeze - out on the final @xmath0-pattern @xcite , especially the formation of the rare - earth peak @xcite , has been investigated extensively in other recent studies . however , so long as @xmath12 equilibrium can be achieved in an @xmath0-process environment , the conditions immediately before the freeze - out in that environment should be close to those derived here . we intend to carry out parametric studies of the @xmath0-process based on more detailed and more realistic astrophysical models in the future , and will explore the effects of various nuclear input on such models . this work was supported in part by the 973 program ( grant no . 2013cb834400 ) , the national natural science foundation of china ( grants no . 10975007 , no . 10975008 , no . 11005069 , no . 11035007 , no . 11105010 , no . 11128510 , no . 11175002 , and no . 11205004 . ) , the research fund for the doctoral program of higher education ( grant no . 20110001110087 ) , the program for new century excellent talents in university ( grant no . ncet-09 - 0031 ) , and the 211 project of anhui university ( grant no . 02303319 - 33190135 ) in the people s republic of china and by the u.s . department of energy under grant no . de - fg02 - 87er40328 at the university of minnesota . j. j. cowan , j. e. lawler , c. sneden , e. a. den hartog , and j. collier , in _ proceedings of the 2006 nasa laboratory astrophysics workshop _ ( nasa / cp-2006 - 214549 ) , edited by v. h. s. kwong and f. s. wreck ( nasa center for aerospace information , hanover , md , 2006 ) , p. 82	we explore the effects of nuclear masses on the temperature and neutron density conditions required for @xmath0-process nucleosynthesis using four nuclear mass models augmented by the latest atomic mass evaluation . for each model we derive the conditions for producing the observed abundance peaks at mass numbers @xmath1 , 130 , and 195 under the waiting - point approximation and further determine the sets of conditions that can best reproduce the @xmath0-process abundance patterns ( @xmath0-patterns ) inferred for the solar system and observed in metal - poor stars of the milky way halo . in broad agreement with previous studies , we find that ( 1 ) the conditions for producing abundance peaks at @xmath1 and 195 tend to be very different , which suggests that , at least for some nuclear mass models , these two peaks are not produced simultaneously ; ( 2 ) the typical conditions required by the critical waiting - point ( cwp ) nuclei with the @xmath2 closed neutron shell overlap significantly with those required by the @xmath3 cwp nuclei , which enables coproduction of abundance peaks at @xmath4 and 195 in accordance with observations of many metal - poor stars ; and ( 3 ) the typical conditions required by the @xmath5 cwp nuclei can reproduce the @xmath0-pattern observed in the metal - poor star hd 122563 , which differs greatly from the solar @xmath0-pattern . we also examine how nuclear mass uncertainties affect the conditions required for the @xmath0-process and identify some key nuclei including @xmath6ni to @xmath7ni , @xmath8zn , @xmath9cd , and @xmath10cd for precise mass measurements at rare - isotope beam facilities .
inflation , the stage of accelerated expansion of the universe , first proposed in the begin of 1980th ( @xcite ) , nowadays attempts a great attention as well . in our previous paper @xcite we investigated the generality of inflation for a wide class of quintessence potentials and noted that criteria we used is weak to decide about the inflationarity of the model with given potential . so in this paper we are proceeding with the investigation of closed friedmann - robertson - walker ( frw ) models , but using power - law potentials only . in this way we calculate spectral indices , slow - roll parameters and other values on the epoch , when the present hubble scale leaves the horizon and at the end of inflation , to compare our constraints with early obtained results and with observation data . this allows us to make some constraints on frw models with scalar field , on potentials of the scalar field and on parameters of these potentials . since due to inflation universe becomes flat very quickly our constraints are applicable to flat case as well . if one supposes that the same potential describes both the inflation stage in early universe and acceleration nowadays we can compare our constraints with constraints on the quintessence potentials ( see e.g. @xcite or @xcite for review of the problem ) . another aim of this paper is the investigation of the influence of the initial conditions on inflation dynamics in case of closed frw models . really , in case of initially flat universe we have only one degree of freedom how does primordial energy density distribute between kinetic and potential terms , but in case of initially closed universe we have at least one more degree of freedom the distribution between curvature and initial expansion rate terms . so our second aim is in studying these two distributions , comparison them to each other and investigation how does they act on inflation . the equations describing the evolution of the universe in closed frw model are @xmath1 @xmath2 and the first integral of the system is @xmath3 where @xmath4 . and we use the _ trigonometrical ( angular ) _ parameterization @xmath5 of the space of initial conditions : @xmath6 @xmath7 @xmath8 @xmath9.\ ] ] our method is as follows . like in @xcite we start from the planck boundary for a given pair of initial conditions @xmath10 and then numerically calculate the further evolution of the universe through inflation . also we calculate scalar spectral index and slow - roll parameters during the evolution of the universe . and since the universe becomes flat very quickly due to the inflation , we can use the usual determination for scalar spectral index . there are two expressions for it first order and second order ones and we use second order results for more precise calculations . also one can express them using _ potential slow - roll approximation _ ( psra ) and _ hubble slow - roll approximation _ ( hsra ) @xcite ( see @xcite for details ) . equations for slow - roll parameters are : for psra @xmath11 @xmath12 for hsra @xmath13 @xmath14 and for @xmath15 respectively : @xmath16 @xmath17 where @xmath18 with euler - mascheroni constant @xmath19 . so we calculate @xmath15 by both these cases and compare them to each other . also we calculate slow - roll parameters and compare them at the epoch when the present hubble scale leaves the horizon for different potentials and for different parameters of the potentials . since the purpose of our paper is to set constraints on inflation models , we compare results of our numerical calculations with constraints on @xmath15 and other values from experiments . in particular we compare them with results obtained in ref . @xcite , so we need to introduce horizon - flow parameters @xmath20 , @xmath21 ( see @xcite for more details ) : @xmath22 @xmath23 . \eqno(5)\ ] ] to compare our constraints with @xcite we need to localize the moment during inflation when the present hubble scale leaves the horizon . it took place at about 60 e - folds before the end of inflation ( @xcite , see also for details ) . this value is model - dependent and it depends also on the way of inflation ends so to simplify we use two values for this 62 as a bound value and 50 . and so we calculate all parameters on these two epochs 62 and 50 e - folds before the end of inflation . for further using we denote this value as @xmath24 . first , let us describe the dependence of the number of e - foldings universe experienced during inflation on initial conditions . to illustrate this we plotted in fig . 1 this dependence on @xmath25 and on @xmath26 separately ; to realize the whole picture one needs to multiply these functions . so in fig . 1(a ) there is a dependence of the number of e - foldings universe experienced during inflation on @xmath25 for negative @xmath25 , in fig . 1(b ) the same but for positive @xmath25 and in fig . 1(c ) on @xmath27 . from fig . 1(a ) and ( b ) one can also see the influence of sign of initial @xmath28 positive @xmath25 corresponds to positive initial @xmath28 and negative @xmath25 corresponds to negative initial @xmath28 . for instance , for @xmath29}$ ] the flat case measure of trajectories experienced insufficient inflation ( the case when universe experienced inflation but the number of e - foldings is less then 70 ) is about @xmath30 for negative @xmath25 and about @xmath31 for positive @xmath25 ( so for flat case the universe experienced sufficient inflation for @xmath32 ; in fig . 1(a ) and 1(b ) dashed line corresponds namely to @xmath33 e - foldings ) . from fig . 1 one can also learn that the dependence of the number of e - foldings on @xmath27 is many times weaker than the dependence on @xmath25 . really , @xmath25 determines initial distribution of the energy between kinetic and potential terms and , since the universe very quickly reaches the slow - roll regime , it determines the energy density at the beginning of inflation . also during the reaching slow - roll regime @xmath26 becomes large , so @xmath27 initial value of @xmath26 weakly acts on the energy density at the beginning of inflation . power - law potentials are well studied and they lead to `` chaotic inflation '' @xcite . one can really use them as inflation part in potentials like those considered by peebles and vilenkin @xcite . they have also attracted attention for some of their properties @xcite . we consider power - law potentials of a kind @xmath34 and our results are as follows . in fig . 2 we plotted positions of the models with different powers on ( @xmath35 ) plane in case of @xmath36 at ( a ) panel and in case of @xmath37 at ( b ) panel . by comparison these plots with bounds on ( @xmath35 ) plane obtained in @xcite from 2df and wmap data one can make a constraint on @xmath38 as @xmath39 in case of @xmath40 and @xmath41 in case of @xmath42 . another constraint on power @xmath38 one can obtain from fig . 3 . in fig . 3 we plotted the dependence of @xmath15 : calculated in psra ( see eq . ( 2 ) ) we denoted as @xmath43}n_s$ ] and calculated in hsra ( see eq . ( 3 ) ) we denoted as @xmath44}n_s$ ] . using bounds on @xmath15 @xcite : @xmath45 ( wmap only ) and @xmath46 ( wmap+acbar+cbi+2dfgrs+@xmath47-forest ) one can set a bound @xmath48 in case of @xmath40 and @xmath49 in case of @xmath42 . one can see these two constraints from @xmath15 and previous one are close to each other . now let us set a constraint on @xmath50 . to do this we can use results obtained from cobe data in @xcite : @xmath51 and we calculate these values at the end of inflation . after defining @xmath52 one can obtain @xmath53 and after substitution ( 6 ) to ( 7 ) : @xmath54 where @xmath55 @xmath56 one needs to keep in mind the relation between @xmath57 and @xmath50 to recalculate @xmath50 into @xmath57 and inversely : @xmath58 another test , also linked with cobe normalization @xcite , is about density perturbation spectrum @xmath59 : @xmath60 and this value is calculated on the epoch when the present hubble scale leaves the horizon . also we can define value of the field on this epoch as @xmath61 and so rewrite ( 8) using ( 6 ) as @xmath62 let us remind the reader that according to cobe data this value is about @xmath63 . using it one can get another estimation for @xmath50 : @xmath64 = 7.5 cm = 12.6 cm finally , in fig . 5 we plotted the dependence of both @xmath65 and @xmath66 on @xmath38 and the dependence of @xmath67 on @xmath38 . the range of possible values of @xmath50 due to eq . ( 7)(@xcite ) is between curves @xmath65 and @xmath66 . we plotted @xmath67 for cases @xmath40 and @xmath42 . one can see from fig . 5 that in case of @xmath68 we have a constraint @xmath41 and in case of @xmath42 we have a constraint @xmath69 . so we reached our aim we set some constraints on power - law potentials and their parameters . we calculated the whole evolution of the universe during inflation for a wide range of initial conditions , parameters of power - law potentials ( power @xmath38 and @xmath57 ) and set some constraints on power - law potentials and on their parameters . also we compared our constraints with results obtained from the recent cosmic microwave background ( cmb ) data and large scale structure ( lss ) data . as we noted above for an epoch when the present hubble scale leaves the horizon we have used two values 62 e - folds before inflation ends and 50 e - folds . and 62 e - folds is bound value in the sense that other possible values are smaller then 62 . we used 50 e - folds namely as an example of such a value and to demonstrate what can happen with @xmath15 , @xmath20 and other values if we use lower ( then 62 ) number of e - folds before the end of inflation to determine the epoch when the present hubble scale leaves the horizon . and the result we obtained is : @xmath0 . the exact value is very model dependent . it depends on many factors , first on the way of inflation ends this determines the epoch when the present hubble scale leaves the horizon . from figures one can see how does it act on the results . also it depends on the observation data to make our constraints more precise we need more precise observation data . but even with these uncertainties our constraints are some harder then results obtained from the cmb and lss data . one can see that our numerical results are some differ from analytical results . this is due to the fact that one can incorrectly determine exact moment when inflation ends using relation @xmath70 . as one can see from fig . 4 ( we plotted it only as an example ; it corresponds to the case @xmath71 ) at _ real _ moment of the end of inflation only @xmath72 is exact equal to unity . and since most analytical results are obtained using relation @xmath73 , it corresponds not to the exact moment of the end of inflation . and the small difference between our numerical results and analytical results is namely due to this uncertainty . this work was supported by the russian ministry of industry , science and technology through the leading scientific school grant @xmath74 2338.2003.2 . we would like to thank a.r . liddle for useful and stimulating discussion and n.yu . savchenko for useful discussion and different help in preparing this paper . guth , phys . d * 23 * , 347 ( 1981 ) ; a.d . linde , phys . 108b * , 389 ( 1982 ) ; a. albrecht and p.j . steinhardt , phys . lett . * 48 * , 1220 ( 1982 ) ; k. sato , mon . not r. astron . * 195 * , 467 ( 1981 ) ; a. starobinsky , pisma a.j . * 4 * , 155 ( 1978 ) [ sov . . lett . * 4 * , 82 ( 1978 ) ] .	we investigate inflation in closed friedmann - robertson - walker universe filled with the scalar field with power - law potential . for a wide range of powers and parameters of the potential we numerically calculate the slow - roll parameters and scalar spectral index at the epoch when present hubble scale leaves the horizon and at the end of inflation . also we compare results of our numerical calculations with recent observation data . this allows us to set a constraint on the power of the potential : @xmath0 .
we consider the coupled rc circuit as shown in fig . [ fig1 ] . two rc circuits of resistances and capacitances @xmath0 and @xmath1 , respectively , are coupled through a third capacitance @xmath2 . the two rc circuits are subject to constant driven currents @xmath3 and @xmath4 , and the voltage differences across the resistors are denoted as @xmath5 and @xmath6 , respectively . the equation of state of this circuit is @xmath7 where @xmath8 , and @xmath9 the two resistors are thermalized at temperature @xmath10 , while voltages across the resistors fluctuate due to the johnson - nyquist ( thermal ) noises @xmath11 and @xmath12 . the noises are assumed to be uncorrelated and gaussian white , and they satisfy the fluctuation - dissipation relation @xmath13 with @xmath14 for @xmath15 , where @xmath16 is boltzmann s constant . via the change of variables @xmath17 , the equation of state can be rewritten as @xmath18 which is mathematically identical to that of a non - driven circuit . the solution of the stochastic equation is @xmath19 in our work we only focus on the steady - state condition , where the first term in eq . [ v_sol ] damps out . using eq . [ thm_fd ] and the fact that @xmath20 is symmetric , time - correlation functions between voltage signals can be derived as @xmath21 and variances and covariances are just their special cases when @xmath22 . using the method of diagonalization , we then derive @xmath23 where @xmath24 and @xmath25 are the eigenvalues of @xmath26 ( @xmath24 is assigned as the larger one ) , @xmath27 , @xmath28 , @xmath29 , and @xmath30 are the @xmath31 element of the matrix @xmath26 . moreover , the correlations between @xmath32 and @xmath33 can be derived as @xmath34 ( no causality ) and @xmath35 . the fokker - planck equation of the full circuit can be shown to be @xmath36 + \frac12 \nabla \cdot \hat{\bf{m}}^{-1 } \hat{\bf{\gamma } } ( \hat{\bf{m}}^{-1})^{t } \nabla p(\vec{v},t)\ , , \ ] ] and the steady - state distribution of @xmath32 is @xmath37 for convenience we use the symbol `` @xmath38 '' to denote that the equality holds up to some normalizing constant that remains invariant in the time - reversal process . note that in the non - driven case the expression reduces to the boltzmann factor of the stored energy in capacitors . the transition probability of the complete description , under an infinitesimal change in time , can be derived as @xmath39\ , , \label{p_f_all}\ ] ] where @xmath40 /dt$ ] , and @xmath33 represents the required noises for such transition . in the corresponding time - reversal process , @xmath41\ , , \ ] ] as one replaces @xmath42 with @xmath43 and @xmath44 with @xmath45 , the latter resulting from inversion of driven currents . subsequently , the net dissipation can be shown to be @xmath46 where @xmath47 represents the currents through the two resistors . the corresponding change in shannon entropy is @xmath48 \nonumber \\ & = 2k_b\ , dt\ , \vec{v}'^t \hat{\bf{\gamma}}^{-1 } \hat{\bf{m } } \dot{\vec{v}}\end{aligned}\ ] ] and the total entropy change @xmath49 is @xmath50 for our current discussions , @xmath51 and @xmath52 are constants , and therefore @xmath53 is a gaussian random variable , while @xmath54 . one can show that @xmath55 it is straightforward of average @xmath56 and width @xmath57 . the corresponding symmetry function is @xmath58 = 2x_0 x/ \sigma^2_x $ ] . thus gaussianity is a sufficient condition of the ft - like behavior . ft is valid if @xmath59 ( in dimensionless units ; in cases where @xmath60 means entropy then ft validates if @xmath61 ) , and for the case where the observed slope is not equal to 1 , ft can be easily restored with the rescaled variable @xmath62 . ] to demonstrate that ft holds for any gaussian random variable whose ratio of variance over mean value is @xmath63 . moreover , with the aid of time - correlation functions , one can also demonstrate the validity of ft over finite - time processes , where @xmath64 , and @xmath65 . in the reduced descriptions , we neglect the signal @xmath6 intentionally , and we would like to check whether ft can still validate with the knowledge of @xmath5 only . note that since the current throught @xmath66 is not measured , the actual dissipation through the resistor is not known , while there exist many methods towards guessing an effective dissipation simply from the time series of @xmath5 . in this work we adopt two methods . in description ( a ) , we treat the time series of @xmath5 as that from a virtual single - rc circuit , and compute the current and therefore dissipation directly following the equation of this simplified circuit . and in description ( b ) , an effective dissipation can be derived using the ratio of forward and backward transition probabilities in @xmath5 over infinitesimal timesteps . we first derive the steady - state probability distribution in @xmath5 : @xmath67 where @xmath68 , and @xmath69 is the effective capacitance . based on eq . [ ss_v1 ] , we can develop a naive interpretation ( `` description ( a ) '' ) , where the masked circuit is treated as a single - rc circuit with capacitance @xmath70 and unmodified @xmath66 ad @xmath3 . thus @xmath71 and @xmath4 are neglected intentionally . this effective single - rc circuit can give the correct steady - state distribution in @xmath5 . alternatively , one can regard the time series of @xmath5 as that from a single - rc circuit , as the effective resistance and capacitance can be derived from its power spectrum , which can be shown to be identical with @xmath66 and @xmath70 , respectively . the total entropy change of this _ gedanken _ single - rc circuit , during an infinitesimal timestep , is @xmath72 where the first term on the rhs represents a `` virtual '' dissipation , as @xmath73 is the virtual current going through @xmath66 in this single - rc circuit . for the case of finite - time difference we have @xmath74 again one finds @xmath75 to be gaussian , while @xmath76 is the average virtual dissipation from @xmath66 . using time - correlation functions , it is straightforward to derive its variance : @xmath77 \bigg\ } \ , . \label{variance_ds1_a}\end{aligned}\ ] ] therefore , ft fails with the adoption of such dissipation function , even at the small-@xmath78 limit . nevertheless , from eq . [ variance_ds1_a ] one finds that this deviation becomes less prominent at large @xmath78 . moreover , one can also show that the deviation from ft diminishes in the weak - coupling regime , as the deviation in variance from @xmath79 is proportional to @xmath80 . beside the above reduced description , one can define the effective dissipation function starting from the forward transition probability of @xmath5 over infinitesimal timesteps . it can be derived by tracing out the degree of freedom in @xmath6 : @xmath81 ^ 2 \right\ } \nonumber \\ & \equiv \exp [ - dt\ , ( \tilde{m } \dot{v_1 } + v'_1)^2 / ( 2 \tilde{\gamma}_1 ) ] \label{p_f_v1}\end{aligned}\ ] ] to the lowest nonvanishing order , where @xmath82 is a two - dimensional vector of elements @xmath83 and @xmath84 . note that this transition probability can be compared to that of a single - rc circuit with the aforementioned effective capacitance @xmath69 , @xmath85 , @xmath86 , and @xmath87 , where @xmath88 . and the renormalized noise amplitude parameter is @xmath89 which is smaller than @xmath90 . since @xmath91 = \exp ( - \tilde{m } v'^{\ , 2}_1 / \tilde{\gamma}_1 ) $ ] , this effective single - rc circuit also gives the correct steady - state probability distribution in @xmath5 . in this effective single - rc circuit , the reversed transition probability is derived simply by replacing @xmath92 with @xmath93 and @xmath94 with @xmath95 in eq . [ p_f_v1 ] . the net dissipation and total entropy change are @xmath96\ , \ \ \text{and } \\ d\tilde{s}^{\rm{(b)}}_{1\rm{tot } } & = d\tilde{s}^{\rm{(b)}}_{1q } - k_b\ln \frac { p_{\rm{ss}}(v_1(t+dt ) ) } { p_{\rm{ss}}(v_1(t ) ) } = 2 k_b \ , dt v_{1d } ( v_1 - \tilde{m } \dot{v}_1 ) / \tilde{\gamma}_1\ , , \label{ds1_b}\end{aligned}\ ] ] respectively . since @xmath97 is linear in @xmath5 and @xmath98 , the total entropy change is gaussian . one can prove that ft is satisfied for @xmath97 . however , violation still occurs in finite - time processes , where @xmath99 and @xmath100 \bigg\ } \ , .\end{aligned}\ ] ] note that , on average , the reduced description ( b ) gives a larger total entropy change than description ( a ) .	in this work we perform theoretical analysis about a coupled rc circuit with constant driven currents . starting from stochastic differential equations , where voltages are subject to thermal noises , we derive time - correlation functions , steady - state distributions and transition probabilities of the system . the validity of the fluctuation theorem ( ft)@xcite is examined for scenarios with complete and incomplete descriptions .
the authors acknowledge some financial support from the spanish project dgicyt - fis2009 - 13364-c02 - 01 . also thanks to the consejera de economa , comercio e innovacin of the junta de extremadura ( spain ) for financial support , project ref . gru09011 .	statistical complexity and fisher - shannon information are calculated in a of quantum scattering , namely the klein tunneling across a potential barrier in graphene . the treatment of electron wave functions as masless dirac fermions us to compute these statistical measures . the comparison of these magnitudes with the transmission coefficient through the barrier is performed . we show that these statistical measures take their minimum values in the situations of total transparency through the barrier , a phenomenon highly anisotropic for the klein in graphene . and statistical indicators ; transmission coefficient ; klein tunneling ; graphene the calculation of entropic measures in quantum bound states @xcite has revealed certain connections with physical properties , such as the ionization potential and the static dipole polarizability @xcite , the closure of shells @xcite and the trace of magic numbers @xcite in atoms and nuclei . the evaluation of these magnitudes for no bound states can have interest in scattering processes . a typical scattering process is the crossing of potential barriers by quantum particles @xcite , a situation which is found in many quantum phenomena , such as tunneling @xcite , interferences @xcite , resonances @xcite , electron transport @xcite , etc . in a previous work @xcite , the relationship of these entropy - information measures with the reflection coefficient in a potential barrier has been investigated . it has been put in evidence that these statistical magnitudes present their minimum values on _ the transparency points _ , just the situations in which the reflection coefficient is null and then the total transmission through the barrier is achieved . following this line , we are also concerned in this work with the calculation of entropic magnitudes in the context of scattering processes through potential barriers , in particular with the so - called klein paradox @xcite . this is an phenomenon with counterintuitive consequences in relativistic quantum mechanics ( rqm ) that has deserved a great interest in particle , nuclear and astro - physics @xcite . from the point of view of the non - relativistic quantum mechanics ( nrqm ) , we know that even if the energy of particles in the incident beam is less than the height of the potential barrier , some of the particles can trespass the barrier ( tunneling effect ) . the higher and wider the potential barrier is , less number of particles can go through the barrier ( exponential decay ) . klein @xcite showed that , in the frame of rqm , it is possible that all particles ( electrons in this case ) can cross the barrier , independently of how high and wide the potential barrier is . so , in this paradoxical context , it is possible to have situations of total transparency . this relativistic effect is explained as a consequence of the matching between electron and positron wavefunctions across the barrier , that allows the tunneling of the electrons . the appearance of positrons inside the barrier can be understood as a result of the sufficiently strong potential of the barrier that is repulsive for the electrons but attractive for the positrons ( holes ) @xcite . the klein paradox has never been realized in laboratory experiments due to the fact that huge electric fields , @xmath0 @xmath1 , are necessary for its observation @xcite . however , a possibility to observe this type of phenomenon that requires lower electric fields has recently been proposed and tested by means of graphene @xcite . graphene is a new material with a promising potential for many technological applications @xcite . it is a monolayer of carbon atoms densely packed in a honeycomb lattice , so it can be considered a two - dimensional ( 2d ) system . in this material , electric fields of order @xmath2 @xmath1 , which are routinely created in realistic samples @xcite , are sufficient to check experimentally the klein tunneling for elementary particles @xcite . from its electronic properties , graphene is a 2d zero - gap semiconductor with low - energy quasiparticles @xcite . these quantum particles can be viewed as massless relativistic fermions that exhibit , at low fermi energies ( @xmath3 ) , a linear dispersion relationship for their energy @xmath4 , @xmath5 where @xmath6 is the planck s constant , @xmath7 is the quasiparticle momentum with @xmath8 the wavevector , and @xmath9 ( instead of @xmath10 for photons ) is the fermi velocity , @xmath11 . these low - energy quasiparticles can be described by the 2d dirac - like hamiltonian given by @xcite @xmath12 where @xmath13 are the pauli matrices . in order to mimic a tunneling experiment in graphene similar to that proposed by klein , a two - dimensional ( @xmath14 plane ) square potential barrier @xmath15 is considered : @xmath16 where @xmath17 and @xmath18 are the height and width of the barrier , respectively , and the dimension along the @xmath19 axis is supposed to be infinite . this local potential barrier can be created by the electric field effect using local chemical doping or by means of a thin insulator @xcite . the effect of this potential barrier is to invert charge carriers inside it , in such a way that holes play the role of positrons , or viceversa . if an incident electron wave propagates under the action of the hamiltonian @xmath20 at an incident angle @xmath21 respect to the @xmath22 axis , the solutions in the different regions for the two - component dirac spinor representing these quasiparticles are : @xmath23 where @xmath24 , @xmath25 are the wavevector components outside the barrier , @xmath26 is the fermi wavevector , @xmath27 is a wavevector component inside the barrier , @xmath28 is the refraction angle , @xmath29 and @xmath30 . the five amplitudes ( @xmath31 ) are complex numbers determined , up to a global phase factor , by the normalization condition and the boundary constraints , namely the continuity of the two components of the dirac spinor at @xmath32 and @xmath33 . the scattering region ( region ii ) provokes a partial reflection of the incident electron wave . the reflection coefficient @xmath34 gives account of the proportion of the incoming electron flux that is reflected by the barrier . the expression for @xmath34 is : @xmath35 where in this case , @xmath36 in this process , there are no sources or sinks of flux , then the transmission coefficient @xmath37 is given by @xmath38 . this coefficient @xmath37 is plotted in fig . [ fig1 ] ( dashed lines ) for two different heights @xmath17 of the potential barrier . observe the anisotropy of its behavior in the sense that apart of the normal incidence ( @xmath39 ) , the total transparency , @xmath40 , can also be found with other angles of incidence . this is just the paradoxical klein tunneling due to the fact that the fermi energy of the incident electrons is much less than the height of the barrier , @xmath41 . this calculation has been performed by taking for the electron and hole concentration , outside and inside the barrier , respectively , the typical values used in experiments with graphene ( see ref . @xcite ) . , in region i and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev.(levels @xmath46 and @xmath47 are plotted in red).,title="fig:",width=264 ] , in region i and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev.(levels @xmath46 and @xmath47 are plotted in red).,title="fig:",width=264 ] \(a ) ( b ) now , we proceed to calculate two statistical magnitudes for this problem , the statistical complexity and the fisher - shannon entropy . these magnitudes are the result of a global calculation done on the probability density @xmath48 given by @xmath49 , taking into account that the region of integration must be adequate to impose the normalization condition in the two - component dirac spinor . as a consequence of having a pure plane wave in the @xmath19 axis , the density is a constant in this direction . then , without loss of generality , if we take a length of unity in the @xmath19 direction , the density does not depend on @xmath50 variable , @xmath51 . the densities obtained for the different regions are : @xmath52 to normalize these densities , the length of the integration interval in the @xmath53 direction is taken to be @xmath54 $ ] , @xmath55 $ ] and @xmath56 $ ] , for regions i , ii and iii , respectively . the statistical complexity @xmath57 @xcite , the so - called @xmath58 complexity , is defined as @xmath59 where @xmath60 is a function of the shannon entropy of the system and @xmath61 tells us how far the distribution is from the equiprobability here , @xmath60 is calculated according to the simple exponential shannon entropy @xmath62 @xcite , that has the expression , @xmath63 with @xmath64 some kind of distance to the equiprobability distribution is taken for the disequilibrium @xmath61 , @xcite , that is , @xmath65 , in region i and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev.,title="fig:",width=264 ] , in region i and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev.,title="fig:",width=264 ] \(a ) ( b ) the fisher - shannon information @xmath66 @xcite is defined as @xmath67 where the first factor is a version of the exponential shannon entropy @xcite , @xmath68 with the constant @xmath69 in the exponential selected to have a non - dimensional @xmath66 . the second factor @xmath70 ^ 2\over \rho(x)}\ ; dx\;,\ ] ] is the so - called fisher information measure @xcite , that quantifies the irregularity of the probability density . the statistical complexity @xmath57 in region i is plotted in solid line in fig . [ fig1 ] . for the points of total transmission , @xmath40 , the dirac spinor is a pure plane wave in the @xmath53 direction due to the fact that there is not reflected wave , @xmath71 in expression ( [ eq - ro1 ] ) , then the density is constant and the complexity @xmath57 is the minimum , @xmath46 . for the values of @xmath21 in between the transparency points , there is reflected wave , @xmath72 in expression ( [ eq - ro1 ] ) , that interferes with the incident one giving rise to standing - like waves . the complexity of any standing wave is @xmath73 , a value that also corresponds to the complexity of the eigenstates of the infinite square well @xcite . observe that the points of total transparency are also well signalled by the minima of the statistical complexity @xmath57 . in fig . [ fig2 ] , the behavior of the fisher - shannon information @xmath66 for the region i is plotted in solid line . the transmission coefficient @xmath37 is also shown in dashed line . similarly to the behavior of @xmath57 in region i , the dirac spinor is a pure plane wave on the points of total transmission , @xmath40 , it means that the density is constant , then the fisher information is null and so @xmath66 . in between the transparency points , standing - like waves are formed due to interference of the reflected waves with incident ones . the fisher information is non null due to the appearance of oscillations in the density , then the fisher - shannon information takes on these regions the maximum values , presenting the biggest one , @xmath74 , for @xmath75 , that is the p value of a perfect standing wave , a fact compatible in this case with a total reflection of the electron flux . , in region ii and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev . ( level @xmath46 is plotted in red).,title="fig:",width=264 ] , in region ii and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev . ( level @xmath46 is plotted in red).,title="fig:",width=264 ] \(a ) ( b ) the behavior of @xmath57 and @xmath66 for the region ii is plotted in solid line in figs . [ fig3 ] and [ fig4 ] , respectively . the transmission coefficient @xmath37 is also shown in dashed line . when the incidence is normal , @xmath39 , the density is a constant as it can be seen in expression ( [ eq - ro2 ] ) , then @xmath57 and @xmath66 take their minimum values , @xmath46 and @xmath76 , respectively . for other angles of incidence , the density presents oscillations , hence @xmath57 and @xmath66 take bigger values than their respective minima . for @xmath75 , there is a situation compatible with a total reflection of the electron wave function that means no penetration in the barrier , then @xmath57 and @xmath66 recover their minimum values . in the region iii , the density is a constant for any angle of incidence , see expression ( [ eq - ro3 ] ) , then @xmath57 and @xmath66 take their minimum values , @xmath46 and @xmath76 . , in region ii and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev.,title="fig:",width=264 ] , in region ii and transmission coefficient , @xmath37 , through a @xmath42 nm . wide barrier vs. the incident angle , @xmath21 , for single - layer graphene . the fermi energy @xmath4 of the incident electrons is taken @xmath43 mev . the barrier heights @xmath17 are ( a ) @xmath44 and ( b ) @xmath45 mev.,title="fig:",width=264 ] \(a ) ( b ) in conclusion , the calculation of the statistical complexity and the fisher - shannon information in an applied problem of electron scattering on a potential barrier in graphene has been presented . the relationship of these indicators with a physical magnitude , the transmission coefficient , has been disclosed . when the transmission through the barrier is complete , these statistical magnitudes take their minimum values , a fact that put in evidence that these entropic measures can also be useful to detect certain physical phenomena in quantum problems , the klein tunneling in graphene included .
recent budding interest on @xmath7 transition metal oxides stems from the promise for emergent novel quantum ground - states resulting from the cooperative effects of strong spin - orbit coupling and electron interactions.@xcite for example , various topological phases are proposed to occur , which include topological insulator , topological semi - metal and quantum spin liquid phases as well as unusual magnetic states and superconductors . in particular , iridates have enjoyed significant attention due to the availability of a variety of materials.@xcite when spin - orbit coupling dominates over crystal field splitting , the basic electronic structure of iridates with ir@xmath3 ions can be described by the pseudospin @xmath8 kramers doublet,@xcite which is a combination of spin and orbital wave functions . hence , in the strong - coupling limit , the ir ions carry @xmath8 moments and the resulting interaction between them presents highly quantum mechanical fluctuations . such an interacting local moment system would be an ideal platform for emergent quantum spin liquid states@xcite when it is placed on geometrically frustrated lattices or the interaction itself has the capacity to generate massive classical degeneracy . currently two different classes of systems have been proposed for possible quantum spin liquid phases in iridates . in the hyperkagome iridate , na@xmath0ir@xmath1o@xmath2,@xcite the ir ions form a three - dimensional network of corner - sharing triangles , which provides the geometric frustration for the heisenberg model . indeed no magnetic ordering has been observed down to a few kelvin in spite of the large curie - weiss temperature , @xmath9.@xcite possible quantum spin liquid and other related phases have been investigated via a variety of theoretical approaches.@xcite in an alternative avenue , 2d ( na@xmath10iro@xmath1 , @xmath11-li@xmath10iro@xmath1 ) honeycomb@xcite and 3d ( @xmath12- or @xmath13-li@xmath10iro@xmath1 ) hyperhoneycomb@xcite iridates have been investigated in the context of bond - dependent interactions , such as the kitaev interaction , between local moments@xcite . such interactions , even if they are placed on bipartite lattices such as the honeycomb lattice , would generate extensive classical degeneracy and may lead to quantum spin liquid phases , as is the case for the pure kitaev model.@xcite in this work , we investigate a generic local - moment model for the hyperkagome iridate , and we examine the effects of anisotropic interactions beyond the heisenberg limit . earlier works on the hyperkagome iridate were focused mostly on the heisenberg model@xcite or the effects of selected sets of anisotropic interactions.@xcite here we provide the derivation of a generic model for @xmath14 local moments of ir@xmath3 ions by taking into account multiorbital interactions such as hund s coupling and spin - orbit coupling . in fact , effects of hund s coupling were not considered in previous studies even though it is crucial for the appearance of anisotropic spin interactions in the hyperkagome iridate that has an edge - sharing structure of iro@xmath15 octahedra.@xcite the missing hund s coupling is taken into account in our microscopic construction of the spin model . remarkably , various frustrating " bond - anisotropic interactions arise . the resulting model is characterized by four parameters ; @xmath16 for the antiferromagnetic heisenberg exchange , @xmath17 for the dzyaloshinskii - moriya interaction , @xmath18 for a kitaev - like term , and @xmath19 for the symmetric anisotropic exchange . the relative sign and form of the three different anisotropic exchange interactions depend on the bond directions and are completely fixed by the lattice symmetries . as a result , the generic model for the hyperkagome iridate has both of the ingredients for massive classical degeneracy , namely geometric frustration and frustrated exchange interactions . indeed all of the @xmath16-only , @xmath18-only , @xmath19-only models are frustrated and support distinct sets of a classically - degenerate manifold . this is in contrast to the honeycomb or hyperhoneycomb systems , in which @xmath16-only model leads to a unique antiferromagnetically - ordered ground - state . given that the curie - weiss temperature is large and negative , we assume that @xmath20 is the dominant energy scale and examine the effects of small anisotropic interactions represented by @xmath17 , @xmath18 , and @xmath19 . this approach is motivated by recent @xmath6sr and nmr experiments in which short - range spin correlations and/or some kind of spin freezing behaviors have been discovered below @xmath21-@xmath22.@xcite the idea is that various small anisotropic interactions become much more important below 6 - 7@xmath18 and the understanding of the nature of the ground - state may require careful examination of the effects of these small perturbations , while the physics of the higher temperature phase may still be understood using the heisenberg interaction . as the first step towards this goal , we map out the magnetic phase diagram for the classical model using the luttinger - tisza@xcite and classical monte carlo simulated annealing methods . in particular , we examine how the massive classical degeneracy of the heisenberg model is lifted depending on which anisotropic interaction is dominant . the resulting phase diagram for a selected set of parameters is shown in fig . [ fig : jkdg ] . there exist three dominant @xmath23 ground - state manifolds with @xmath24 or two distinct @xmath25 discrete degeneracies , labelled as @xmath26 and @xmath27 , in addition to an incommensurate magnetic order . here @xmath26 ( @xmath27 ) refers to the manifold of classical ground - states where the direction of one ( two ) of the moments on each triangle is almost parallel to the local @xmath28 axis at each site , which we will define later , while the remaining two ( one ) are not ( see figs . [ fig : z6_2p ] and [ fig : z6_1p ] ) . when @xmath29 and @xmath30 , the @xmath23 state with @xmath24 degeneracy ( which is called the canted windmill state@xcite ) is the ground - state . if @xmath29 and @xmath31 , the @xmath23 state with @xmath26 degeneracy is dominant for relatively large @xmath19 . on the other hand , when @xmath32 and @xmath31 , the @xmath23 state with @xmath27 degeneracy becomes dominant for large @xmath18 . in general , @xmath33 promote the @xmath23 states with @xmath34 degeneracy , respectively ( see the phase diagrams in figs . [ fig : windmill ] , [ fig : z6_2p ] , and [ fig : z6_1p ] ) . if the @xmath23 states with discrete degeneracy dominate the low temperature short - range spin correlations , one of the degenerate states , once it is formed locally in certain regions , may not easily relax to another degenerate configurations . this is due to the constrained spin dynamics in frustrated magnets , as explained in the main text . the energy / temperature scale where these phenomena occur will be set by the dominant anisotropic interactions . this may explain the spin freezing or slow spin dynamics observed in the experiments . our results also suggest that it would be fruitful to investigate quantum spin liquid phases that may be obtained by quantum disordering the @xmath23 states described above . . site classification and local @xmath28 axes . the table lists the site type and @xmath28 axis for the 12 sites in a unit cell ( described in fig . [ fig : lattice ] ) . [ tab : localc2 ] [ cols="^,^,^,^,^,^",options="header " , ] for a simple setting of the strong - coupling expansion , we consider a two - site system described by the following hamiltonian . @xmath35 we assume that each site has five electrons ( or one hole ) in the @xmath5 manifold as in the ir@xmath3 ion . the above hamiltonian consists of the atomic multiorbital interactions ( @xmath36 ) , spin - orbit coupling ( @xmath37 ) , and electron hoppings ( @xmath38 ) between the two sites . in the strong - coupling expansion , we assume that @xmath39 . first , we consider the system in the atomic limit described by only the interaction term : @xmath40 here , we employed the well known kanamori hamiltonian for the multiorbital interactions . the hamiltonian is parametrized with the intraorbital coulomb interaction ( @xmath41 ) and the hund s coupling ( @xmath42 ) . it is readily diagonalized in terms of the total hole number ( @xmath43 ) in the @xmath5 manifold , and total spin ( @xmath44 ) and total orbital ( @xmath45 ) angular momenta at each site ( @xmath46 ) . hence , the eigenstates at each site can be represented by the three quantum numbers : @xmath47 . notice that we are using the hole basis here instead of the electron basis . we summarize the eigenstates with @xmath48 in table [ tab : kanamori_eigen ] . the atomic ground - state manifold is characterized with the quantum numbers : @xmath49 . next , we turn on the atomic spin - orbit coupling : @xmath50 note that the coupling constant ( @xmath51 ) has a minus sign when written in the hole basis . we incorporate effect of the spin - orbit coupling by projecting the atomic ground - state manifold @xmath52 into the @xmath53 kramers doublet ( where @xmath54 ) . we introduce electron / hole hoppings between the two sites . we assume the most generic hopping hamiltonian as follows . @xmath55 here , @xmath56 are the hole annihilation operators at the site @xmath57 . the subscripts @xmath58 represent the single - hole ( @xmath59 ) states in the @xmath5 basis , and @xmath60 means the spin state of the hole . the hopping amplitude matrix @xmath61 is parametrized with nine independent real parameters : @xmath62 the hopping matrix is basically decomposed into the trace ( @xmath63 ) , antisymmetric vector ( @xmath64 ) , and traceless symmetric matrix ( @xmath65 ) . now the effective exchange interactions are derived with the strong - coupling expansion . as mentioned earlier , we assume that @xmath66 . by reflecting the hopping effects on the @xmath53 doublets via the second order perturbation theory , we obtain the effective exchange interactions : @xmath67 it must be noted that here the operators @xmath68 are the @xmath53 pseudospin operators at the sites , 1 and 2 . the coupling constants are given by the following expressions . [ eq : jdg ] @xmath69 , \\ \tilde{\bm{d } } & = & -\frac{16}{9 } \left [ \left(\frac{2}{u-3j_h } + \frac{1}{u+2j_h}\right ) s\bm{v } + \frac{j_h \bm{q } \bm{v } } { ( u-3j_h)(u - j_h)}\right ] , \\ \tilde{\bm{\gamma } } & = & \frac{4}{27 } \left [ \left(\frac{15}{u-3j_h } + \frac{1}{u -j_h } + \frac{8}{u+2j_h}\right ) \left(\bm{v}\bm{v}^{\mathrm{t}}-\frac{\bm{v}^2}{3}\right ) + \frac{18j_h ( \bm{q}^2 -\frac{1}{3}\mathrm{tr}(\bm{q}^2 ) ) } { ( u-3j_h)(u - j_h ) } \right].\end{aligned}\ ] ] one can easily check that the @xmath70 matrix is traceless : @xmath71 . ) in the idealized crystal structure . here each ir - o bond is parallel to one of the global @xmath72 axes and the bond length is uniform across all the bonds . the o@xmath73 ( black ) and o@xmath10 ( gray ) sites are distinguished by different local environment : o@xmath73 ( o@xmath10 ) sites are shared by neighboring three ( two ) ir ions . for comparison , the actual crystal structure of na@xmath0ir@xmath1o@xmath2@xcite is drawn together with faint gray balls . , width=264 ] we construct the @xmath53 spin model for the hyperkagome iridate na@xmath0ir@xmath1o@xmath2 . in the actual crystal structure of na@xmath0ir@xmath1o@xmath2,@xcite most of the anisotropic couplings in @xmath74 [ eq . ( [ eq : hamiltonian_2 ] ) ] are expected to be nonzero due to lattice distortions from an ideal structure . in this work , instead of pursuing the actual crystal structure , we idealize the structure in such a way that each ir - o bond is parallel to one of the global @xmath72 axes and the bond length is uniform across all the bonds ( fig . [ fig : iro6 ] ) . with the idealized crystal structure , we derive a relatively simple , but still generic spin hamiltonian for na@xmath0ir@xmath1o@xmath2 . we show the derivation by taking the ir - ir bond ( 6,5 ) in fig . [ fig : lattice ] as an example . first , we express the hopping matrix at the bond by using the slater - koster parametrization:@xcite @xmath75 here , the parameters @xmath76 , @xmath77 , @xmath78 represent direct hoppings between the neighboring ir sites . the parameter @xmath79 means the hoppings between the @xmath80 orbitals at the o@xmath81 site and @xmath5 orbitals at an ir site , and @xmath82 implies the energy difference between the former and latter orbitals . hence , the amplitude @xmath83 indicates the indirect hopping via the intermediate oxygen site o@xmath81 . notice that there are two distinct oxygen sites , o@xmath73 and o@xmath10 , in the compound ( see fig . [ fig : iro6 ] ) . the o@xmath73 ( o@xmath10 ) sites are shared by neighboring three ( two ) ir ions . we assume that @xmath84 and @xmath85 for the inequivalent o@xmath73 and o@xmath10 sites . the above hopping matrix can be decomposed into the form of eq . ( [ eq : hopping_mat ] ) with the following nonzero parameters : [ eq : svq ] @xmath86 the resulting exchange interactions caused by the above hopping amplitudes take the following form . @xmath87 microscopic expression for the coupling constants @xmath88 , @xmath89 , @xmath90 and @xmath91 can be obtained by plugging eq . ( [ eq : svq ] ) into ( [ eq : jdg ] ) . lastly , we simplify the above bond hamiltonian into the final form : @xmath92 here , the compling constants for the heisenberg ( @xmath16 ) , kitaev ( @xmath18 ) , dzyaloshinskii - moriya ( @xmath17 ) , and anisotropic & symmetric ( @xmath19 ) interactions are defined as follows . @xmath93 the exchange interactions at other bonds are generated by applying the @xmath94 and @xmath28 symmetry operations ( sec . [ sec : lattice ] ) to the bond hamiltonian @xmath95 . then , we obtain the model hamiltonian @xmath96 [ eq . ( [ eq : hamiltonian ] ) ] . the luttinger - tisza analysis ( lta ) for the @xmath16-@xmath18 model is discussed in details here . first , the lta is briefly reviewed.@xcite in the lta , we relax the hard spin constraint @xmath97 and implement it on average : @xmath98 ( @xmath99 is the number of the spin moments ) . the resulting quadratic hamiltonian matrix is solved in the momentum space . @xmath100 in this expression , the @xmath101 matrix @xmath102 is the block hamiltonian matrix in the momentum @xmath65 sector ( @xmath103 is the number of sublattices in a unit cell , and @xmath104 in our hyperkagome lattice model ) . the @xmath105-component column vector @xmath106 represents a fourier component of real - space spins @xmath107 . after finding the lowest - energy state of @xmath96 , we check whether the state satisfies the hard spin constraint . if it does , the lowest - energy state is the _ exact _ ground - state of the hamiltonian . when the hard spin constraint is not satisfied , the lta provides a lower bound for the ground - state energy . now we apply the lta to the @xmath16-@xmath18 model @xmath108 . one can easily find that the lowest - energy state occur at @xmath109 by diagonalizing the hamiltonian matrix @xmath110 ( see fig . [ fig : lta - jk ] ) . hence , we focus on the @xmath109 sector of @xmath108 and analyze the spin structure of the lowest - energy mode . the @xmath109 hamiltonian can be block - diagonalized in the following way : @xmath111 here , we take the basis for the spin vector as follows . @xmath112 the three 12@xmath11312 matrices @xmath114 are given by @xmath115 with the submatrices @xmath116 here , we use the reduced coupling constant @xmath117 . notice that @xmath114 are equivalent matrices connected by unitary transformations . the twelve energy eigenvalues shared by the three matrices are obtained as follows . * @xmath1181 * 0 ( 2-fold ) * @xmath119 ( 2-fold ) * @xmath120 ( 2-fold ) * @xmath121 * @xmath122 * @xmath123 * @xmath124 as shown in fig . [ fig : lta - jk ] ( d ) , a different lowest - energy state is selected by the kitaev interaction depending on the sign of @xmath125 . when the kitaev interaction is antiferromagnetic ( @xmath126 ) , the ground - state has the energy @xmath127 and the corresponding spin state is constructed in the following way . first , we note that the three matrices @xmath114 have the same eigenvector : @xmath128 with @xmath129 . then , we obtain the eight degenerate spin states by linearly combining the @xmath130 as @xmath131 with the sign factors @xmath132 . one can check that the eight states satisfy the hard spin constraint . these states are the eightfold - degenerate ferrimagnetic states mentioned in the main text ( fig . [ fig : z6_2p ] ) . in the case of the ferromagnetic kitaev interaction ( @xmath133 ) , the matrices @xmath114 have almost the same eigenvector except for the sign structure : @xmath134 , @xmath135 , @xmath136 with @xmath137 . the ground - state manifold can be constructed by taking eight different combinations of @xmath130 , as we did in eq . ( [ eq : linearcomb ] ) . one can check the hard spin constraint for each of the eight states , and find that the ground - state manifold consists of the @xmath24 windmill states and @xmath138 states ( fig . [ fig : z6_2p ] ) . in this case , the ground - state energy is given by @xmath139 . model with ( a ) @xmath140 , ( b ) @xmath141 , and @xmath142.,width=302 ] in this appendix , we describe the lta for the @xmath16-@xmath19 model @xmath143 . the lowest - energy state occurs at different positions in the brillouin zone depending on the value of the coupling constant @xmath19 ( see fig . [ fig : lta - jg ] ) . we focus on the @xmath144 case in which the lta provides the exact ground - states of @xmath143 . in this case , the lowest - energy mode has a flat dispersion along the @xmath19r line in the brillouin zone . however , it turns out that none of the finite-@xmath65 states satisfies the hard spin constraint ( the absence of the finite-@xmath65 states is also confirmed by our simulated annealing approaches ) . in the following , we will examine the @xmath109 states and construct the ground - state manifold for the @xmath144 case . interestingly , the @xmath109 states are the lowest - energy states for each of @xmath145 and @xmath146 . the @xmath109 states of @xmath143 can be obtained by investigating the ground - state manifold of @xmath146 and then considering the heisenberg interaction on the manifold . hence , we solve the @xmath19-only model @xmath146 first . we set @xmath147 and block diagonalize the hamiltonian matrix in the following fashion . @xmath148 where @xmath149 is the 9@xmath1139 matrix @xmath150 with @xmath151 here , as the basis for the spin vector we choose @xmath152 with the following sequence of the spin components . @xmath153 the lowest eigenvalue of @xmath154 is @xmath155 with the doubly degenerate eigenvectors , @xmath156 and @xmath157 . the ground - states satisfying the hard spin constraint can be constructed by combining the eigenvectors in the following ways . @xmath158 in this expression , the right hand side shows three different ways for the combinations with the sign factors @xmath159 . therefore , we find 3@xmath1132@xmath160=48 different states in the ground - state manifold of @xmath146 . to obtain the ground - states of @xmath143 , the heisenberg interaction is considered on the ground - state manifold of @xmath146 . by examining the heisenberg interaction energy for each state , one can find that only six states in the manifold have the minimum energy , @xmath161 per bond . the six states represented by @xmath162 are the @xmath163 states ( fig . [ fig : z6_1p ] ) . the above three pairs of vectors sequently correspond to @xmath163 @xmath164 states , respectively . it is interesting to note that the state vectors do not depend on the coupling constant @xmath19 . wind - mill state on the ( a ) @xmath165 , ( b ) @xmath166 and ( c ) @xmath167 planes.,width=264 ] -@xmath168 state on the ( a ) @xmath165 , ( b ) @xmath166 and ( c ) @xmath167 planes . , width=264 ] -@xmath169 state on the ( a ) @xmath165 , ( b ) @xmath166 and ( c ) @xmath167 planes . , width=264 ] here we provide the static spin structure factors for the @xmath23 magnetic orders . these structure factors are computed for the long - range ordered cases . for each @xmath23 order , the structure factor is calculated with the formula @xmath170 here , @xmath99 is the number of spin moments and @xmath171 represents the real space position of the moment @xmath44 at site @xmath57 . note that for any @xmath23 state the fourier component of @xmath44 is non - vanishing only when the wave vector @xmath65 is equal to a reciprocal lattice vector . accordingly , the structure factor has nonzero peaks only at the reciprocal lattice vectors . the structure factors for the @xmath23 orders are plotted in figs . [ fig : sf_z2 ] , [ fig : sf_z62p ] and [ fig : sf_z61p ] , where @xmath172 , with @xmath173 being integers . in these figures , one can see that the structure factor is zero at @xmath109 for all three states.@xcite it is also seen that for the @xmath24 windmill state , the structure factor patterns for all three planes are the same . this results from the @xmath94 rotational invariance of the @xmath24 state . on the other hand , for the @xmath174-@xmath168 and @xmath175-@xmath169 states , the structure factor pattern on the @xmath176 plane is different from the others , because of the broken @xmath94 rotational symmetry . moreover , @xmath174-@xmath168 and @xmath175-@xmath169 states show different patterns in the structure factor . these differences may be used to distinguish three different @xmath109 states . in this appendix , we provide a quantitative estimation of the energy barrier between members of the six degenerate @xmath138/@xmath163 states . the direct path to move from one member to another in the spin configuration space is obtained by rotating the spins one - by - one until the spin - configuration reaches another member . in the estimation of the energy barrier , we rotate spins in random order and compute the energy per site measured from the ground - state energy as a function of the number of spin rotations . figure [ fig : eb ] shows the energy barrier for several pairs of the @xmath25 states as a function of the number of spin rotations for various system sizes . we can see that the energy barrier has a peak at @xmath177 ( i.e. , when half of the spins are rotated ) . we can also see that the curves for different system sizes are scaled into a single curve . this indicates that the energy barrier is proportional to the system size . therefore , if a short - range correlation is formed in a relatively large region , the energy cost to overcome the barrier is very large . notice that this energy barrier arises due to the fact that the spin configuration is taken out of the ground - state manifold of the unperturbed heisenberg model during the spin rotation processes . k. a. modic , t. e. smidt , i. kimchi , n. p. breznay , a. biffin , s. choi , r. d. johnson , r. coldea , p. watkins - curry , g. t. mccandless , j. y. chan , f. gandara , z. islam , a. vishwanath , a. shekhter , r. d. mcdonald , and j. g. analytis , nat . commun . * 5 * , 4203 ( 2014 ) .	the hyperkagome iridate , na@xmath0ir@xmath1o@xmath2 , has been regarded as a promising candidate material for a three - dimensional quantum spin liquid . here the three - dimensional network of corner - sharing triangles forms the hyperkagome lattice of ir@xmath3 ions . due to strong spin - orbit coupling , the local moments of ir@xmath3 ions are described by the pseudospin @xmath4 kramers doublet . the heisenberg model on this lattice is highly frustrated and quantum / classical versions have been studied in earlier literature . in this work , we derive a generic local - moment model beyond the heisenberg limit for the hyperkagome iridate by considering multi - orbital interactions for all the @xmath5 orbitals and spin - orbit coupling . the lifting of massive classical degeneracy in the heisenberg model by various spin - anisotropy terms is investigated at the classical level and the resulting phase diagram is presented . we find that different anisotropy terms prefer distinct classes of magnetically ordered phases , often with various discrete degeneracy . the implications of our results for recent @xmath6sr and nmr experiments on this material and possible quantum spin liquid phases are discussed .
a dynamic phase transition ( dpt ) , in which the dynamical behavior of a nonequilibrium system changes in a singular way at a critical value of a system parameter , can provide insight into the often complex behavior of such systems . dpts of various kinds have been identified and studied in chemical , @xcite charge - density wave , @xcite and superconducting systems . @xcite the dpt of interest here was first identified in computer simulations of magnetic systems . @xcite when a bistable magnetic system in its ferromagnetic phase is exposed to an alternating magnetic field , the response depends strongly on the period of the applied field . at low values of the field period , the system effectively can not respond to the rapidly changing field , and its magnetization oscillates in a restricted range around one of its two nonzero ( magnetized ) values . at high values of the period , the magnetization can follow the field , resulting in a square , symmetric hysteresis loop . this behavior suggests the cycle - averaged magnetization q ( with a value near @xmath5 at low field period , and near 0 at high field period ) as a candidate for a ` dynamic ' order parameter . it has indeed been shown computationally @xcite that at a critical period @xmath6 , there exists a second - order phase transition with respect to @xmath3 , with critical exponents consistent with those of the equilibrium ising transition . for further description of this dpt and the numerical evidence for the second - order phase transition , see refs . . also associated with the appearance of a non - zero value of @xmath3 below the critical period @xmath6 is a significant decrease in the hystersis - loop area . since first being identified , this dpt has received a great deal of attention in numerical simulations @xcite and analytical work . @xcite however , to date there have been only tentative experimental indications of its presence , principally the collapse of the hystersis loop area with decreasing field period , in several studies of hysteresis scaling properties in magnetic thin films.@xcite it is important to understand the extent to which this dpt , which has been well studied computationally , can be realized in an experimental system . to achieve this , one must investigate the quantitative behavior of the dynamic order parameter itself . to this end , we identified an experimental system , an ultra - thin [ co(4 )/pt(7 )]@xmath1-multilayer , which possesses strong perpedicular anisotropy and which exhibits substantial similarities @xcite to a two - dimensional kinetic ising model , @xcite for which the dpt is well - documented . @xcite we present here the first quantititative data on the behavior of the dynamic order parameter in this ising - like experimental system , and we compare it to the behavior observed in kinetic monte carlo ( mc ) simulations . recent computational work @xcite by our group on the two - dimensional kinetic ising model identified the cycle - averaged magnetic field , @xmath2 , as a significant component of the field conjugate to @xmath3 , and established the existence of a fluctuation - dissipation relation ( fdr ) between the susceptibility @xmath7 and the variance @xmath8 in the vicinity of the critical period . applying those results here , we use the variance @xmath9 as a proxy for the susceptibility @xmath7 in our comparisons of experimental and computational results . in order to establish the presence of a phase transition unequivocally , it is necessary to demonstrate power - law scaling with well - defined critical exponents . while this often can be done straightforwardly enough in computational work , experimentally it requires precise measurements at carefully controlled values of the relevant thermodynamic variables . this was achieved for the equilibrium curie transition in an experimental thin film only fairly recently.@xcite in the present case of a [ co / pt]@xmath1 multilayer driven by an alternating magnetic field , the nonlinear , hysteretic nature of the electromagnet used to generate the alternating field resulted in small fluctuations of the cycle - averaged field , @xmath2 , rendering it impossible to study the experimental system at @xmath10 , as done in all but the most recent @xcite computational studies . moreover , it was not possible to establish scaling relations with statistical significance from our data , despite achieving an applied field and bias field amplitude precision of 0.5% and 0.1% , respectively , in our measurements . nonetheless , our data for @xmath3 and @xmath4 are consistent with power - law scaling near the critical point . in spite of this , we argue that the similarity of the behavior of the cycle - averaged magnetization @xmath3 in the multilayer to its behavior in the kinetic ising model , coupled with an explanation of the differences which do exist in terms of known physical properties of the multilayer , provides strong evidence for the presence of this dpt in the multilayer . in addition , we propose refinements in the experimental technique which may produce in the future the precision required to investigate power - law scaling . however , peaks in response functions also consitute evidence for phase transitions , and our observation of a peak in the variance @xmath4 represents significantly stronger evidence for this dpt in an experimental system than that previously published.@xcite the importance of the present study is twofold . first , it provides new insight into the dynamic behavior of an ultrathin magnetic film system whose properties are important to the development of future generations of ultrahigh density information - recording technologies . second , it provides valuable experimental input to theorists efforts to develop a comprehensive understanding of phase transitions in far - from - equililibrium systems . this paper is organized as follows . in section [ s : expt ] , we describe the experimental setup and procedure . in section [ s : results ] , we present our experimental results , consisting of directly measured hysteresis loops and magnetization time series , as well as nonequilibrium phase diagrams ( nepds ) used to characterize the behavior of the dynamic order parameter and its fluctuations . in section [ s : comparison.discussion ] , we compare the experimental results to those of computer simulations of a kinetic ising model and argue that the similarities provide strong evidence for the presence of the dpt in the multilayer system . in section [ s : conclusion ] we present our conclusions , as well as suggestions for further experimental work . the ultra - thin [ co(4 )/pt(7 )]@xmath1-multilayer samples were prepared by low - pressure ( 3 mtorr ar ) magnetron sputtering onto ambient temperature silicon - nitride coated si substrates . we deposited 200 pt as a seed layer , and the samples were coated with 20 pt to avoid contamination . these ferromagnetic multilayers have an easy axis along the surface normal and strong uniaxial anisotropy , resulting in high remanent magnetization and square out - of - plane hysteresis loops . @xcite x - ray diffraction confirmed a ( 111 ) crystalline texture , with a lateral crystallographic coherence length from several tens to several hundreds of nm . due to the strong ferromagnetic interlayer coupling , mediated by the pt between adjacent co layers , @xcite the entire multilayer acts as a single magnetic film with strong uniaxial anisotropy . while out - of - plane magnetized ferromagnetic films tend to form equilibrium domain structures to reduce magnetostatic interactions , in ultra - thin films the energy reduction resulting from domain formation is extremely small , so that this effect is strongly suppressed . @xcite we measured the effective anisotropy field of the multilayer to be @xmath11 koe , more than an order of magnitude larger than the fields used in our experiment . the spins should therefore remain strongly collinear perpendicular to the film . these experimental facts , combined with theoretical @xcite and experimental @xcite evidence that ultrathin films with uniaxial anisotropy are in the same universality class as the equilibrium ising model , suggest that the two - dimensional kinetic ising model can serve as a useful model for the _ nonequilibrium _ behavior of our multilayer . to study magnetization reversal in the multilayer , we measured the polar magneto - optical kerr effect ( moke ) , which is proportional to the magnetization along the surface normal . a time - dependent magnetic field along the surface normal was provided by an electromagnet . in addition , for reasons discussed in sec . [ s : results ] , a small constant additional magnetic field - the ` bias field ' , @xmath2 - was applied by separate means . the total ( time - varying ) field actually experienced by the multilayer was monitored by a hall probe . data were recorded for two multilayer samples , a and b. the electromagnet was driven by a computer - controlled bi - polar power supply to provide a saw - tooth field sequence for several values of the period between @xmath12 s and @xmath13 s ( sample b : between @xmath14 s and @xmath15 s ) . for each sequence , we first measured two complete , saturated hysteresis loops using a large field amplitude @xmath16 oe ( for both samples a and b ) . next , in the actual measurement sequence , the amplitude was lowered to @xmath17 oe ( sample b : @xmath18 oe ) , and 49 identical cycles were run . by monitoring the field sequence , we determined that the variation of @xmath19 during the 49 cycles was less than 1% . finally , the amplitude was increased back to @xmath20 to trace out another complete hysteresis loop . the two previous experiments@xcite which observed hysteresis loop collapse with decreasing field period were conducted on 3 monolayer ( 3 ml ) co / cu(001 ) ultrathin films with in - plane uniaxial anisotropy @xcite and on 2 - 3 ml fe / w(110 ) ultrathin films with perpendicular uniaxial anisotropy.@xcite these experiments also employed an electromagnet to generate the time - varying magnetic field , a hall probe to record the magnetic field at the location of the sample , and the moke effect to measure the sample magnetization . in our experiment , however , two additional steps were taken to enable a more thorough investigation of the experimental behavior of the cycle - averaged magnetization , @xmath21 previously identified as the order parameter for the dpt in the kinetic ising model.@xcite ( the index @xmath22 in eq . ( [ dpt.def ] ) denotes the number of the field cycle . ) first , as mentioned above , a significant component of the field conjugate to the dynamic order parameter was identified in a recent computational study @xcite as the cycle - averaged value of the magnetic field , or ` bias field ' , @xmath23 in ref . , with @xmath2 taken as a small constant field superposed on a time - varying square - wave field , the conjugate field scaling exponent @xmath24 in the dpt was shown to agree with the equilibrium ising field scaling exponent @xmath25 to within computational error . the bias field was found to have a signficant effect on the dynamic order parameter @xmath3 , especially near the critical period where the susceptibility @xmath7 has its maximum . in a computer simulation , one can study the dpt in precisely zero bias field without difficulty . in experiment , however , due to nonlinearities in the electromagnet used to generate the time - varying field , the actual bias field @xmath2 experienced by the sample will inevitably exhibit finite values , which may even fluctuate slightly during the measurement sequence . we therefore chose to run experiments at a series of different values of the bias field , which were measured and controlled carefully using the hall probe,@xcite at each value of the field period studied . second , a recent study of thicker [ co / pt]@xmath26 multilayers @xcite found that hard - to - reverse , residual bubble domains could persist beyond an apparent saturation field , and could serve as nucleation sites during magnetization reversal when the field was subsequentely reversed . in the experiments reported here , the two complete loops at high saturating field before the measurement sequence ensured that the samples began each run in a well defined magnetic state , and that the effects of residual bubble domains at the lower field magnitude of the measurement cycle were consistent across different data runs . in addition , comparison of the complete loops before and after the measurement sequence served to verify the stability of our experimental setup during each run . the hysteresis loops and magnetization time series in samples a and b had very similar characteristics , so we show only those of sample a ( the nepds for both samples a and b will be presented , however ) . figure [ mh.data ] shows the normalized magnetization , @xmath27 , vs @xmath28 for periods @xmath29 s and @xmath30 s in sample a. the initial and final complete loops exhibit full remanent magnetization _ in a single - domain state _ , and a sharp reversal region . the loops in the measurement sequence , at field magnitude @xmath19 , are also square with a sharp reversal region . however , a sharp suppression of the magnetization reversal process is visible below a pinning field @xmath31 oe ( sample b : @xmath32 oe).@xcite the pinning phenomenon has been studied quantitatively in multilayer films , and it is understood to result from a distribution of energy barriers to domain wall motion resulting from lattice defects and variations in the film thickness . @xcite the kerr microscope image in fig . [ kerr.image ] illustrates the domain reversal pattern in a sister sample , at magnetization @xmath33 after nucleation from a positive saturated state . it shows clearly that , above the pinning field @xmath34 , the reversal within our 1.0 mm@xmath35 moke laser spot occurs by a multidroplet ( md ) process @xcite of nucleation and growth of approximately circular domains . this reversal pattern is similar to that previously observed in ultrathin , single - layer copt films,@xcite and can be contrasted with the reversal pattern in thicker [ co / pt]@xmath26 multilayers,@xcite in which larger magnetostatic interactions lead to the formation of stripe domains . the md process is quite similar to that occurring near the dpt in the kinetic ising model , @xcite indicating that , if the effect of the pinning field is properly accounted for in the analysis , the kinetic ising model can serve as a reasonable model of the multilayer . as described in section [ s : expt ] , given the important role of the bias field @xmath2 as shown in ref . , we have studied the behavior ( the average value and the variance ) of @xmath3 in the multilayer with respect to both @xmath0 and @xmath2 , and compared this to the corresponding behavior of @xmath3 in the two - dimensional kinetic ising model . the two previous experimental reports of the collapse of the hysteresis loop with decreasing field period concentrated principally on the decrease of the hysteresis loop area ( cf . fig . 1 of ref . , and figs . 1 and 2 of ref . ) . the cycle - averaged magnetization was also plotted , in fig . 3 of ref . , where its magnitude was shown to become nonzero and then to increase as the applied field amplitude is decreased at a constant value of the field period . however , the bias field , which has a strong effect on the cycle - averaged magnetization near the dynamic phase transition in computational studies @xcite was not carefully controlled , and the variance in the cycle - averaged magnetization was not recorded . from an initial value of @xmath36 oe ( sample b : @xmath37 oe ) , the bias field , as generated separately from the time - varying field of the electromagnet , was increased in steps of 0.5 oe ( sample b : steps of @xmath38 oe ) . the actual value of the bias field during the 49 measurement cyles was determined , using data from the hall probe , to fluctuate at the 0.1% level of the applied field amplitude , @xmath19 . specifically , the actual bias field could be characterized by a confidence interval @xmath39 , with @xmath40 oe . in fig . [ mt.data.experiment](a ) , magnetization time series for sample a at @xmath29 s are shown in a strong negative , weak negative , and weak positive bias field . the corresponding time series , @xmath41 , of the cycle - averaged magnetization are plotted in fig . [ mt.data.experiment](b ) . for both negative @xmath2 values , the cycle - averaged value of the magnetization ( and in fact of any quantity ) settles eventually into a non - equilibrium steady state ( ness ) , so that @xmath41 is observed in fig . [ mt.data.experiment](b ) to fluctuate around an average value @xmath42 . ( the fluctuations of @xmath41 in the ness arise from variations in the minor loop amplitude , as seen in fig.[mh.data](a ) , which are caused principally by thermally - induced fluctuations in the extent and timing of nucleation events . ) the transition to the ness with @xmath42 takes longer in the weak negative @xmath2 than in the strong negative @xmath2 , and it does not occur at all in the weak positive @xmath2 . at the longer period @xmath30 s , in contrast , the time series adjust quickly to a ness with a small @xmath43 value of the same sign as @xmath2 , as shown in figs . [ mt.data.experiment](c ) and [ mt.data.experiment](d ) . this behavior will be compared to that in the kinetic ising model in the next section . to characterize the behavior of the cycle - averaged magnetization in the multilayer more fully , we measured its average @xmath43 over the field cycles for which the system was in a ness , for a range of values of @xmath2 at each period @xmath0 . we will refer to the resulting contour plot of @xmath43 vs @xmath0 and @xmath2 , shown in fig . [ nepd.samplea](a ) , as a non - equilibrium phase diagram ( nepd ) , so named in analogy to plots of thermodynamic relationships in equilibrium phase diagrams . as described in section [ s : intro ] , the recent confirmation @xcite of a fdr near the critical point of the dpt in the two - dimensional kinetic ising model justifies the use of the variance as a proxy for the susceptibility in evaluating evidence for the dpt . the nepd of the variance @xmath9 within the ness is shown in fig . [ nepd.samplea](b ) . the corresponding nepds for sample b are presented in fig . [ nepd.sampleb ] . to determine in which field cycle the system should be considered to have entered a ness , we used the following procedure . the first 10 field cycles were discarded to minimize initial transient effects . we then identified the first subsequent field cycle , @xmath44 , for which the slope of the remaining values @xmath45 could not be statistically distinguished from zero . if more than 15 values of @xmath41 remained , the mean @xmath43 and variance @xmath8 of these values were calculated ; otherwise , no data were recorded on the nepds for that pair of values @xmath46 . this procedure ensured that , for example for the measurement sequences in strong negative and weak negative bias fields in fig . [ mt.data.experiment](a ) , the field cycles making up a transition to the ness with @xmath47 were not included as part of the respective ness s . in fig . [ nepd.samplea](a ) , the boundary between regions with negative @xmath43 and those with positive @xmath43 occurs , for the lower periods , at a slightly negative bias field , @xmath48 oe . similarly , the maximum of @xmath9 occurs , for the lower periods , at a slightly negative bias field , @xmath49 oe . this appears to contradict the expected @xmath50 inversion symmetry of a ferromagnetic system , which predicts that the nepds should be symmetric with respect to the bias field @xmath2 . however , it can be explained by the presence of residual , hard - to - reverse bubble domains , with coercivities @xmath51 in the range @xmath52 , as observed recently in thicker [ co / pt]@xmath26 multilayers.@xcite since the final saturating field before our measurement sequence begins is positive , such residual domains ( if present ) would remain positively magnetized during the measurement sequence , in which the field magnitudes @xmath53 would be too weak to reverse them . the bubble domains would then serve as nucleation centers for reversal on the increasing branch of the hysteresis loops during the measurement sequence , lowering the nucleation field on this branch . this effect can indeed be observed in fig . [ illustrate.persistent](a ) , in which the nucleation field @xmath54 on the increasing branch ( from the negatively magnetized state ) in the final five measurement cycles is seen to be smaller than the nucleation field in the increasing branch of the cycles at saturation field @xmath20 . in contrast , the nucleation field @xmath55 on the decreasing branch ( from the positive magnetized state ) in the first five measurement cycles is approximately equal to the nucleation field in the decreasing branch of the cycles at saturated field @xmath20 . this asymmtery in the nucleation fields has an effect on the value of @xmath3 in equal and opposite bias fields , @xmath2 and @xmath56 , at a given field period in sample a. in the positive bias field @xmath2 , the reversal window beginning from positive saturation consists of the field intervals @xmath57 and @xmath58 , whereas in the negative bias field @xmath56 , the reversal window beginning from negative saturation consists of the field intervals @xmath59 and @xmath60 . since @xmath61 , the magnetization reversal proceeds farther in the bias field @xmath56 than in the bias field @xmath62 , leading to an asymmetry in the values of the cycle - average magnetization @xmath3 , with @xmath63 . thus , the boundary between the regions with negative @xmath43 and positive @xmath43 is shifted from @xmath10 oe to the slightly negative bias field @xmath64 oe . we now compare the experimental results to the behavior of the two - dimensional kinetic ising model with nearest - neighbor ferromagnetic interactions , for which the dpt has been conclusively established , @xcite under conditions similar to those of the experiments . the energy of the ising system is given by @xmath65 with @xmath66 the exchange constant , @xmath67 the time - dependent magnetic field , and @xmath68 the spin magnetic moment at site @xmath22 . we use the glauber transition rule , @xmath69 , with the next spin to be updated chosen at random . having verified that larger lattice sizes do not change the results appreciably , we use a @xmath70 square lattice with periodic boundary conditions . the curie temperature of the multilayer was estimated from ref . as @xmath71 k. we therefore choose the simulation temperature @xmath72 , to correspond to the temperature @xmath73 of the experiments . as in ref . , the reversal dynamics are found to be md for @xmath74 at this temperature , with mean reversal time @xmath75 monte carlo steps per spin ( mcss ) . for a sawtooth waveform with amplitude @xmath74 , we determine the dpt by finite - size scaling analysis @xcite to occur at @xmath76 mcss . to recreate the experimental conditions as closely as possible , we simulated magnetization time series consisting of two saturated loops at @xmath77 , followed by 50 field cycles at @xmath74 . the period values were selected from both below and above @xmath6 , from @xmath78 to @xmath79 mcss . at each value of @xmath0 , we collected magnetization time series for a range of bias fields from @xmath80 to @xmath81 . figures [ mt.data.simulation](a ) and [ mt.data.simulation](b ) show simulated magnetization time series , and the corresponding time series of the cycle - averaged magnetization , @xmath41 , for a strong negative , weak negative , and strong positive bias field at period @xmath82 mcss , just below @xmath83 mcss . as in the experimental time series in figs . [ mt.data.experiment](a ) and [ mt.data.experiment](b ) , the transition to the ness with @xmath42 in the simulated time series takes longer in the weak negative @xmath2 than in the strong negative @xmath2 , and it does not occur at all in the weak positive @xmath2 . in addition , the behavior of the simulation in strong negative and weak positive @xmath2 at the higher period @xmath84 mcss , shown in figs . [ mt.data.simulation](c ) and [ mt.data.simulation](d ) , resembles the behavior of the experimental system at higher period shown in figs . [ mt.data.experiment](c ) and [ mt.data.experiment](d ) . the larger splitting between the average values of @xmath43 in fig . [ mt.data.experiment](d ) , relative to fig . [ mt.data.simulation](d ) , is due mainly to the effect of the pinning field in the experimental system , which results in an increased sensitivity to the bias field , as discussed later in this section . the similarity of the simulation data in fig . [ mt.data.simulation ] to the experimental data in fig . [ mt.data.experiment ] suggests that the experimental system may be near criticality just above @xmath85 s , but this similarity does not by itself provide conclusive evidence of a dpt . to evaluate the question more thoroughly , we constructed nepds of @xmath86 and @xmath8 , measured in the ness in simulation , with respect to @xmath0 and @xmath2 . two procedures were used to generate simulation data in a ness . in the first procedure , we mimicked the experimental data analysis precisely , by identifying the part of the simulated time series which consituted the ness using the same criterion described in section [ s : results ] . in the second procedure , the simulation was initialized , for a run at particular values @xmath46 , such that the final saturation field before the measurement sequence was of the same sign as @xmath2 . this field - symmetric initial condition begins each simulation near the ness , for both positive and negative @xmath2 , thus eliminating the transitions seen in fig . [ mt.data.experiment](a ) . we then ran 32 independent mc realizations of 50 field cycles each , including the final 40 cycles of each realization in the ness data . the simulated nepds of @xmath86 and @xmath4 using the first procedure are shown in figs . [ nepd.sim.singlerun](a ) and [ nepd.sim.singlerun](b ) , respectively , while those produced using the second procedure are shown in figs . [ nepd.sim.steadystate](a ) and [ nepd.sim.steadystate](b ) . we concentrate first on the comparison of simulation data to the data from experimental sample a , for reasons discussed below . the simulated nepds of fig . [ nepd.sim.singlerun ] are clearly similar in form to the experimental nepds in fig . [ nepd.samplea ] . however , there are two significant differences , which can be accounted for by the presence in sample a of hard - to - reverse , residual bubble domains , and of a pinning field for domain - wall motion , respectively . the first difference is in the position of the peak of the variance @xmath4 . in the experimental nepd , the peak of @xmath4 is situated at ( @xmath87 s@xmath88 oe ) , whereas in the simulated nepd , the ( somewhat diffuse ) peak occurs at periods just above @xmath89 mcss . while one might argue from fig . [ nepd.sim.singlerun ] that the peak is centered around a slightly negative @xmath2 , a collection of nepds ( produced by different mc realizations ) showed no net tendency toward either positive or negative @xmath2 . as described at the end of section [ s : results ] , the shift in the boundary between the regions of positive and negative @xmath43 values , as well as the shift of the experimental peak of @xmath4 , toward negative @xmath2 , can be explained by the asymmetry in nucleation fields during the measurement sequence . this asymmetry is in turn due to the presence of the residual bubble domains , which were magnetized positively by the final ( positive ) saturating field before the beginning of the measurement sequence . the second principal difference between the simulated and experimental nepds is the gradual slope of the contour lines extending from the left side of the experimental nepd in fig . [ nepd.samplea](a ) , as compared to the steep drop of the corresponding contours in the simulated nepd in fig . [ nepd.sim.singlerun](a ) . if one normalizes the @xmath2 values ( i.e. , the @xmath90 axis ) by the field magnitude @xmath19 in both figs . [ nepd.samplea](a ) and [ nepd.sim.singlerun](a ) , this difference in slopes becomes even more pronounced . physically , the difference in slopes corresponds to a susceptibility @xmath91 which falls off more slowly with increasing period in the experiment than in the simulation . this increased sensitivity to the bias field results from the pinning field in the multilayer . in the experiment , the entire magnetization reversal takes place within the restricted field intervals @xmath58 and @xmath92 , whereas in the simulated reversal it occurs , albeit with a domain wall velocity decreasing linearly with the applied field , @xcite within the entire field intervals @xmath93 and @xmath94 . a given bias field , expressed as a percentage of @xmath19 , thus constitutes a larger percentage of the field interval during which magnetization reversal occurs in experiment than in the simulation , and so has a larger effect on @xmath43 . when the second data analysis procedure was used , including averaging over 32 independent simulation runs , the nepd of @xmath43 , shown in fig . [ nepd.sim.steadystate](a ) , became symmetric with respect to @xmath2 , both above and below @xmath6 . in addition , the nepd of @xmath9 in fig . [ nepd.sim.steadystate](b ) became more sharply focused near @xmath10 at periods just above @xmath6 . this strongly suggests that the asymmetry with respect to @xmath2 in fig . [ nepd.samplea](a ) , and size of the peak region in fig . [ nepd.samplea](b ) , would decrease if the second procedure described above were followed in gathering and analyzing experimental data . in comparing the simulation data to the results for experimental sample b , we note that in the nepd of @xmath43 for sample b in fig . [ nepd.sampleb](a ) , there is much less ( if any ) shift of the boundary between the region of positive @xmath43 and the region of negative @xmath43 toward negative bias field . similarly , the peak of @xmath8 in fig.[nepd.sampleb](b ) is just slightly further from @xmath10 oe than the uncertainty @xmath95 oe . the very minimal difference between the nucleation field in the increasing saturating fields and that in the increasing measurement loops in fig . [ illustrate.persistent](b ) , as compared to the sizable difference noted above for sample a in fig . [ illustrate.persistent](a ) , suggests strongly that there were significantly fewer impurities and/or variations in film thickness giving rise to residual bubble domains in sample b. the nepds of @xmath43 and @xmath8 for sample b closely resemble the part of the simulated nepds ( fig . [ nepd.sim.singlerun ] ) above @xmath96 mcss ( @xmath97 mcss ) . thus , we tentatively place the location of the critical period at ( @xmath14 s , @xmath98 oe ) , i.e. , at the left edge of the nepds , as indicated in fig . [ nepd.sampleb ] . this is also consistent with the fact that the magnetization reversal reaches @xmath99 for sample b at @xmath100 s in fig . [ illustrate.persistent](b ) , whereas it reaches only @xmath101 for sample a at @xmath12 s in fig . [ illustrate.persistent](a ) . in simulations , it has been observed @xcite that the dpt occurs at a period where the magnetization reversal ( from saturation ) proceeds at least as far as @xmath102 . since the data from sample a includes periods from both below and above our critical period , @xmath103 s , with the variance @xmath4 visibly decreasing in fig . [ nepd.samplea](b ) both below and above this period , we have focused on sample a as our primary evidence for the observation of the dpt . the similarity of the experimental results from sample a to the simulated results , with respect to ( i ) the behavior of the time series @xmath41 in various @xmath2 , in figs . [ mt.data.experiment ] and [ mt.data.simulation ] , ( ii ) the form of the nepds of @xmath43 , in figs . [ nepd.samplea](a ) and [ nepd.sim.singlerun](a ) , and ( iii ) most importantly , the existence of a well - defined peak in the nepd of @xmath4 in fig . [ nepd.samplea](b ) ; constitute strong evidence for this dpt in an experimental magnetic system . the differences between the experimental and simulated nepds of @xmath86 and @xmath4 can be consistently explained , for both samples a and b , in terms of the effect of hard - to - reverse bubble domains in sample a , as described in section [ s : results ] , and the presence of a pinning field for domain - wall motion , as described earlier in this section . as mentioned in section [ s : intro ] , the kinetic ising model we have used does not give rise to a pinning field , and thus presents a simplified model of the motion of domain walls in the [ co / pt]@xmath1-multilayer . significant progress has been made in using a preisach - arrhenius ( pa ) model of magnetic viscosity to describe the thermally activated motion of domain walls over a distribution of free - energy barriers associated with variations in film thickness , grain boundaries , and impurities.@xcite the pa model , while quite useful to describe phenomenologically the decay of magnetization on long time scales , does not directly incorporate the spatial dependence and cooperative nature of magnetization processes . in addition , recent work @xcite has shown that the dependence of the energy barriers ( to wall motion ) on the applied field deviates from the linear dependence @xcite usually assumed in pa models . it appears likely that incorporating both spatially varying magnetization , such as that occurring in the md reversal process ( cf . fig [ kerr.image ] ) , and a realistic description of a wide distribution of energy barriers will require a computationally intensive , multi - scale micromagnetic model , for which the dpt would first have to be established computationally . we have therefore chosen to model the spatially varying md reversal process faithfully with a two - dimensional kinetic ising model , and to account for magnetic viscosity effects ( i.e. , the pinning field ) in our analysis and comparison to experiment . the differences between the results of the experiments and the kinetic ising simulations are consistent with what one would expect from the added effects of magnetic viscosity in an experimental multilayer system exhibiting the dpt . we have compared the behavior of the cycle - averaged magnetization , @xmath3 , in experiments on a [ co / pt]@xmath1 multilayer , to its behavior in mc simulations of the two - dimensional kinetic ising model , in which a dpt with respect to the period @xmath0 of an applied alternating field has been well established . @xcite plots of time series of the magnetization ( and of @xmath3 ) in the presence of various bias fields @xmath2 , as well as non - equilibrium phase diagrams of the average , @xmath43 , and the variance , @xmath4 , as functions of @xmath0 and @xmath2 , were used to characterize and compare the behavior of @xmath3 in experiment and simulation . the behavior was seen to be similar , with differences that could be clearly accounted for in terms of the known phenomena of a pinning field and residual bubble domains . the results , in particular the presence of a clear peak in the variance @xmath4 in experimental sample a , provide the strongest experimental evidence to date for the presence of this dpt in a magnetic system . furthermore , the results strongly suggest that the cycle - averaged magnetization @xmath3 serves as a dynamic order parameter in the [ co / pt]@xmath1-multilayer system . it would be most desirable to perform further and more precise experiments on the dpt in perpendicular magnetic ultrathin films . specifically , it would be useful to collect data using the second procedure described in section [ s : comparison.discussion ] , i.e. , to ensure that the final saturated state before the measurement cycle has the same sign as the applied bias field . also , one could attempt to apply a field magnitude @xmath19 that is significantly larger than the pinning field @xmath34 , thereby forcing the critical period to lower values . one would expect the effect of the pinning field , for example in creating more gradual contours in the non - equilibrium phase diagram of figs . [ nepd.samplea](a ) and [ nepd.sim.singlerun](a ) , to be reduced in this case . however , this will be experimentally very challenging given the hysteretic nature and response time of the electromagnet one would need in this case to generate the time - varying applied field sequences . in addition , a lower value of the critical period would enable the collection of data for more field cycles in each ness . as noted above , some level of experimental fluctuations in the bias field is probably unavoidable . however , with a sufficiently large number of cycles recorded in a ness , one could attempt to calculate values of @xmath43 and @xmath4 using data from a subset of the recorded cycles with a more restricted range of @xmath2 values . this could potentially supply the precision required to investigate power - law scaling near the critical period . the added statistics provided by forcing the dpt to occur at a shorter period , as well as more precise control over the temporal stability of the bias field , could enable extraction of the critical exponents from the power - law behavior of @xmath4 vs @xmath104 and @xmath2 . finally , in an ultrathin au / co / au film it was observed @xcite that a mixture of nucleation at fixed ( ` soft ' , extrinsic ) sites and at random ( intrinsic ) sites occurred . if this also occurs in [ co / pt ] multilayers , it could affect both the size of the fluctuations of the dynamic order parameter and its spatial correlations , relative to those in the kinetic ising system , in which nucleation occurs only at random sites . one could examine a series of real - time images of the sample , such as the one shown in fig . [ kerr.image ] , in order to calculate spatial and time - displaced correlation functions to address this question . the results from such additional experiments could serve to further reinforce the evidence for the dpt presented here . to achieve an even higher degree of confidence , one would need to formulate an accurate , multiscale micromagnetic model of the multilayer , and to demonstrate the existence of the dpt in this computational model . given the large system sizes needed for computational finite - size scaling analysis , such an effort will likely require innovations in simulation methods , as well as further increases in available computing power . the computational work was supported by nsf grant nos . dmr-0120310 and dmr-0444051 , by the center for computational sciences at mississippi state university , and by the school of computational science at florida state university . in addition , one of us ( d.r . ) acknowledges support from clarkson university and the use of resources under nsf grant no . another author ( a. b. ) would like to acknowledge funding from the department of industry , trade and tourism of the basque government and the provincial council of gipuzkoa under the etortek program , project ie06 - 172 , as well as from the spanish ministry of science and education under the consolider - ingenio 2010 program , project csd2006 - 53 . the kinetic ising model does simplify certain aspects of the magnetization dynamics in the multilayer , mainly the dependence of the domain - wall velocity on the applied field . this issue is discussed in detail in section [ s : comparison.discussion ] . the hall probe was calibrated accurately by determining the reading of the hall probe corresponding to true zero field at the sample as the point at which the cycle - averaged magnetization changed from negative to positive value at high field period . the pinning field was determined by first averaging over a collection of increasing ( decreasing ) @xmath50 curves , such as the increasing curves near @xmath105 in fig . [ mh.data](a ) . the widest field range @xmath106 for which all @xmath27 values outside the field range were greater ( smaller ) than the average value , @xmath107 , within the field range was then found . finally , the endpoints @xmath108 and @xmath109 from eight such procedures were used to calculate the mean value and uncertainty of @xmath34 given in the text . despite applied field magnitudes and bias field magnitudes precise to the 0.5% and 0.1% level , respectively , of the applied field magnitude , it was not possible to establish scaling relations with statistical significance from our data . experimental data of the normalized magnetization , @xmath27 , vs field @xmath28 for multilayer sample a. data were taken for two initial saturated loops , 49 loops at amplitude @xmath110 koe , and one final saturated loop . ( a ) period @xmath29 s. ( b ) @xmath30 s. in both cases , the bias field , which is defined and discussed in section [ s : results ] , was @xmath111 oe . , title="fig:",width=288 ] experimental data of the normalized magnetization , @xmath27 , vs field @xmath28 for multilayer sample a. data were taken for two initial saturated loops , 49 loops at amplitude @xmath110 koe , and one final saturated loop . ( a ) period @xmath29 s. ( b ) @xmath30 s. in both cases , the bias field , which is defined and discussed in section [ s : results ] , was @xmath111 oe . , title="fig:",width=288 ] kerr microscope image of magnetization in a sister [ co / pt]@xmath1 multilayer sample , taken shortly after the start of magnetization reversal ( at @xmath112 ) . the area represented in the image had physical dimensions 0.88 @xmath113 0.62 mm@xmath35.,width=384 ] ( color online ) . time series of the normalized magnetization , @xmath27 , and of the cycle - averaged magnetization , @xmath41 , in multilayer sample a , at two different values of the period . ( a ) period @xmath29 s , magnetization time series . ( b ) @xmath29 s , @xmath41 vs measurement cycle index @xmath22 . ( c ) @xmath30 s , magnetization time series . ( d ) @xmath30 s , @xmath41 vs measurement cycle index @xmath22 . note that for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted ; however , the full time series of @xmath41 ( calculated from the 49 measurement cycles only ) are shown . the bias field values corresponding to strong negative , weak negative , and weak positive are @xmath114 and @xmath115 oe , respectively.,title="fig:",width=288 ] ( color online ) . time series of the normalized magnetization , @xmath27 , and of the cycle - averaged magnetization , @xmath41 , in multilayer sample a , at two different values of the period . ( a ) period @xmath29 s , magnetization time series . ( b ) @xmath29 s , @xmath41 vs measurement cycle index @xmath22 . ( c ) @xmath30 s , magnetization time series . ( d ) @xmath30 s , @xmath41 vs measurement cycle index @xmath22 . note that for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted ; however , the full time series of @xmath41 ( calculated from the 49 measurement cycles only ) are shown . the bias field values corresponding to strong negative , weak negative , and weak positive are @xmath114 and @xmath115 oe , respectively.,title="fig:",width=288 ] ( color online ) . time series of the normalized magnetization , @xmath27 , and of the cycle - averaged magnetization , @xmath41 , in multilayer sample a , at two different values of the period . ( a ) period @xmath29 s , magnetization time series . ( b ) @xmath29 s , @xmath41 vs measurement cycle index @xmath22 . ( c ) @xmath30 s , magnetization time series . ( d ) @xmath30 s , @xmath41 vs measurement cycle index @xmath22 . note that for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted ; however , the full time series of @xmath41 ( calculated from the 49 measurement cycles only ) are shown . the bias field values corresponding to strong negative , weak negative , and weak positive are @xmath114 and @xmath115 oe , respectively.,title="fig:",width=288 ] ( color online ) . time series of the normalized magnetization , @xmath27 , and of the cycle - averaged magnetization , @xmath41 , in multilayer sample a , at two different values of the period . ( a ) period @xmath29 s , magnetization time series . ( b ) @xmath29 s , @xmath41 vs measurement cycle index @xmath22 . ( c ) @xmath30 s , magnetization time series . ( d ) @xmath30 s , @xmath41 vs measurement cycle index @xmath22 . note that for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted ; however , the full time series of @xmath41 ( calculated from the 49 measurement cycles only ) are shown . the bias field values corresponding to strong negative , weak negative , and weak positive are @xmath114 and @xmath115 oe , respectively.,title="fig:",width=288 ] ( color online ) . non - equilibrium phase diagrams ( nepds ) for sample a , showing ( a ) the average @xmath43 and ( b ) the variance @xmath116 in the non - equilibrium steady state ( ness ) , as functions of the period @xmath0 and bias field @xmath2 . the portion of the magnetization time series which consituted the ness was determined for each ( @xmath117 ) according to the procedure described in section [ s : results ] . within each time series , the bias field fluctuated within a range @xmath118 , with @xmath40 oe as shown . the black dot shows the estimated location of the critical point of the dpt.,title="fig:",width=288 ] ( color online ) . non - equilibrium phase diagrams ( nepds ) for sample a , showing ( a ) the average @xmath43 and ( b ) the variance @xmath116 in the non - equilibrium steady state ( ness ) , as functions of the period @xmath0 and bias field @xmath2 . the portion of the magnetization time series which consituted the ness was determined for each ( @xmath117 ) according to the procedure described in section [ s : results ] . within each time series , the bias field fluctuated within a range @xmath118 , with @xmath40 oe as shown . the black dot shows the estimated location of the critical point of the dpt.,title="fig:",width=288 ] ( color online ) . normalized magnetization , @xmath27 , vs field , @xmath28 , in ( a ) sample a at ( @xmath119 s , @xmath120 oe ) , and ( b ) sample b at ( @xmath14 s , @xmath121 oe ) . the upper thin lines ( blue ) show the first 5 measurement cycles , which reach ( nominal ) positive saturation , while the lower thin lines ( red ) show the final 5 measurement cycles , which reach ( nominal ) negative saturation . the thick black lines show two complete cycles at the saturation field , @xmath122 koe . the nucleation fields in the measurement and saturated cycles correspond closely , except for the increasing branch ( from negative saturation ) in ( a ) , due to the presence of positively magnetized residual bubble domains in sample a ( see text ) . as the @xmath123 axis range was reduced to more clearly show the nucleation fields , the saturation loops continue outside the field range shown.,title="fig:",width=288 ] ( color online ) . normalized magnetization , @xmath27 , vs field , @xmath28 , in ( a ) sample a at ( @xmath119 s , @xmath120 oe ) , and ( b ) sample b at ( @xmath14 s , @xmath121 oe ) . the upper thin lines ( blue ) show the first 5 measurement cycles , which reach ( nominal ) positive saturation , while the lower thin lines ( red ) show the final 5 measurement cycles , which reach ( nominal ) negative saturation . the thick black lines show two complete cycles at the saturation field , @xmath122 koe . the nucleation fields in the measurement and saturated cycles correspond closely , except for the increasing branch ( from negative saturation ) in ( a ) , due to the presence of positively magnetized residual bubble domains in sample a ( see text ) . as the @xmath123 axis range was reduced to more clearly show the nucleation fields , the saturation loops continue outside the field range shown.,title="fig:",width=288 ] ( color online ) . time series of normalized magnetization , @xmath27 , and of cycle - averaged magnetization , @xmath41 , from simulations of the @xmath124 kinetic ising model , at two different values of the period and in various bias fields . ( a ) period @xmath125 mcss , magnetization time series . ( b ) @xmath125 mcss , @xmath41 vs measurement cycle @xmath22 . ( c ) @xmath84 mcss , magnetization time series . ( d ) @xmath84 mcss , @xmath41 vs measurement cycle @xmath22 . as in fig . [ mt.data.experiment ] , for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted . however , the full time series of @xmath41 , calculated from the 50 measurement cycles , are shown . the bias fields values corresponding to strong negative , weak negative , and weak positive are @xmath126 , @xmath127 , and @xmath128.,title="fig:",width=288 ] ( color online ) . time series of normalized magnetization , @xmath27 , and of cycle - averaged magnetization , @xmath41 , from simulations of the @xmath124 kinetic ising model , at two different values of the period and in various bias fields . ( a ) period @xmath125 mcss , magnetization time series . ( b ) @xmath125 mcss , @xmath41 vs measurement cycle @xmath22 . ( c ) @xmath84 mcss , magnetization time series . ( d ) @xmath84 mcss , @xmath41 vs measurement cycle @xmath22 . as in fig . [ mt.data.experiment ] , for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted . however , the full time series of @xmath41 , calculated from the 50 measurement cycles , are shown . the bias fields values corresponding to strong negative , weak negative , and weak positive are @xmath126 , @xmath127 , and @xmath128.,title="fig:",width=288 ] ( color online ) . time series of normalized magnetization , @xmath27 , and of cycle - averaged magnetization , @xmath41 , from simulations of the @xmath124 kinetic ising model , at two different values of the period and in various bias fields . ( a ) period @xmath125 mcss , magnetization time series . ( b ) @xmath125 mcss , @xmath41 vs measurement cycle @xmath22 . ( c ) @xmath84 mcss , magnetization time series . ( d ) @xmath84 mcss , @xmath41 vs measurement cycle @xmath22 . as in fig . [ mt.data.experiment ] , for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted . however , the full time series of @xmath41 , calculated from the 50 measurement cycles , are shown . the bias fields values corresponding to strong negative , weak negative , and weak positive are @xmath126 , @xmath127 , and @xmath128.,title="fig:",width=288 ] ( color online ) . time series of normalized magnetization , @xmath27 , and of cycle - averaged magnetization , @xmath41 , from simulations of the @xmath124 kinetic ising model , at two different values of the period and in various bias fields . ( a ) period @xmath125 mcss , magnetization time series . ( b ) @xmath125 mcss , @xmath41 vs measurement cycle @xmath22 . ( c ) @xmath84 mcss , magnetization time series . ( d ) @xmath84 mcss , @xmath41 vs measurement cycle @xmath22 . as in fig . [ mt.data.experiment ] , for clarity , only the two initial saturated cycles and the first 26 measurement cycles of the magnetization time series are plotted . however , the full time series of @xmath41 , calculated from the 50 measurement cycles , are shown . the bias fields values corresponding to strong negative , weak negative , and weak positive are @xmath126 , @xmath127 , and @xmath128.,title="fig:",width=288 ] ( color online ) . non - equilibrium phase diagrams ( nepds ) for the @xmath124 kinetic ising model , with parameters given in section [ s : comparison.discussion ] , of ( a ) @xmath43 and ( b ) @xmath116 in a non - equilibrium steady state ( ness ) , as functions of the period @xmath0 and bias field @xmath2 . the data for the ness were drawn from a single run of 50 field cycles , using the same analysis which was employed for the ( experimental ) figs . [ nepd.samplea ] and [ nepd.sampleb ] and described in section [ s : results ] . the black dot shows the location of the critical point , @xmath129 mcss @xmath130 , as determined by finite - size scaling analysis.,title="fig:",width=288 ] ( color online ) . non - equilibrium phase diagrams ( nepds ) for the @xmath124 kinetic ising model , with parameters given in section [ s : comparison.discussion ] , of ( a ) @xmath43 and ( b ) @xmath116 in a non - equilibrium steady state ( ness ) , as functions of the period @xmath0 and bias field @xmath2 . the data for the ness were drawn from a single run of 50 field cycles , using the same analysis which was employed for the ( experimental ) figs . [ nepd.samplea ] and [ nepd.sampleb ] and described in section [ s : results ] . the black dot shows the location of the critical point , @xmath129 mcss @xmath130 , as determined by finite - size scaling analysis.,title="fig:",width=288 ] ( color online ) . same as fig . [ nepd.sim.singlerun ] , but using a different procedure to generate the data in the non - equilibrium steady state ( ness ) . here , the data for the ness were drawn from the final 40 cycles and averaged over 32 independent mc simulations , each of which was initialized in a saturation field of the same sign as its associated @xmath2 . , title="fig:",width=288 ] ( color online ) . same as fig . [ nepd.sim.singlerun ] , but using a different procedure to generate the data in the non - equilibrium steady state ( ness ) . here , the data for the ness were drawn from the final 40 cycles and averaged over 32 independent mc simulations , each of which was initialized in a saturation field of the same sign as its associated @xmath2 . , title="fig:",width=288 ]	a dynamic phase transition ( dpt ) with respect to the period @xmath0 of an applied alternating magnetic field has been observed previously in numerical simulations of magnetic systems . however , experimental evidence for this dpt has thus far been limited to qualitative observations of hysteresis loop collapse in studies of hysteresis loop area scaling . here , we present significantly stronger evidence for the experimental observation of this dpt , in a [ co(4 )/pt(7 )]@xmath1-multilayer system with strong perpendicular anisotropy . we applied an out - of - plane , time - varying ( sawtooth ) field to the [ co / pt]@xmath1 multilayer , in the presence of a small additional constant field , @xmath2 . we then measured the resulting out - of - plane magnetization time series to produce nonequilibrium phase diagrams ( nepds ) of the cycle - averaged magnetization , @xmath3 , and its variance , @xmath4 , as functions of @xmath0 and @xmath2 . the experimental nepds are found to strongly resemble those calculated from simulations of a kinetic ising model under analagous conditions . the similarity of the experimental and simulated nepds , in particular the presence of a localized peak in the variance @xmath4 in the experimental results , constitutes strong evidence for the presence of this dpt in our magnetic multilayer samples . technical challenges related to the hysteretic nature and response time of the electromagnet used to generate the time - varying applied field precluded us from extracting meaningful critical scaling exponents from the current data . however , based on our results , we propose refinements to the experimental procedure which could potentially enable the determination of critical exponents in the future .
clusters of galaxies are ideal tracers of the largest density fluctuations in the universe , and their abundance ( and its evolution with cosmic time ) may be used to place constraints on cosmology ( e.g. , @xcite ) . they also provide ideal laboratories for studying galaxy evolution . originally used in this way since they contain a large number of galaxies all in the same location , it has since become clear that the properties of their member galaxies are markedly different from galaxies in the general field ( e.g. , @xcite , @xcite , @xcite ) , implying that mechanisms which truncate star formation and transform galaxy morphology operate on cluster scales ( e.g. , @xcite and references therein ) . constructing large , well - defined samples of galaxy clusters has a long and varied history . the first systematic searches involved visual identification of overdensities of optical galaxies on photographic plates @xcite . in the 1970s , with the advent of x - ray telescopes above the earth s atmosphere , selection of clusters from their extended x - ray emission found favour @xcite . recently , a combination of large format ccd detectors and objective algorithms to search efficiently for signatures of galaxy clusters has led to a revival in the use of optical selection in cluster surveys @xcite . a variety of techniques have been suggested to exploit the expected luminosity and/or color distribution of galaxies in clusters . the big advantage of these surveys compared with the older visual searches is that the detection method could be automated and characterised , meaning that the survey selection function could be quantified . arguably the most efficient method is that of @xcite which uses the fact that the cores of galaxy clusters are dominated by galaxies with old stellar populations , forming a tight red - sequence in color magnitude space @xcite . a number of other realisations of red - sequence based cluster finding algorithms exist ( e.g. , @xcite ) differing in some details but all relying on accurate colors from imaging in two or more filters . the observed color of this sequence provides an accurate distance estimate . the application of this method led to the construction of the first red - sequence cluster survey ( rcs-1 , @xcite ) , a 72 square degree imaging survey in two bands ( @xmath11 and @xmath12 ) designed to locate galaxy clusters from @xmath13 using the technique of @xcite . not only is optical selection of galaxy clusters undergoing a revival , but astronomy in general is entering an era of ` survey science ' where an unprecedented number of wide - field optical ( and nir ) surveys are currently underway or planned , such as lsst @xcite ; pan - starrs ; ukidss @xcite ; and des . in addition , many of these wide - field optical surveys are specifically targeted at areas surveyed for clusters using other methods , such as the blanco cosmology survey @xcite of the south pole telescope ( spt , @xcite ) sunyaev zeldovich effect ( sz)-selected cluster survey . for surveys using other methods ( such as sz selection ) , the optical data are critical for the verification of the cluster candidates found and for the determination of photometric redshifts . furthermore , surveying the same areas with multiple techniques allows important comparisons of the different selection effects and the resulting properties of the clusters found ( e.g. , @xcite ) . in this paper we describe the second red - sequence cluster survey ( rcs-2 ) , the largest survey of this new generation for which imaging has already been completed . this builds on the methodology of rcs-1 . the rcs collaboration has invested a large amount of work in attempting to characterize the selection function and the properties of clusters selected with this technique . many of these results are directly applicable to rcs-2 ( such as mass richness calibrations ) and so it is useful to summarize some of the rcs work to date . the efficiency of the selection method employed by the red - sequence surveys is that it can locate and estimate the redshifts of clusters using only one color ( two filter ) data , given the appropriate choice of filters . it is impractical to obtain mass estimates from follow - up observations of the @xmath030 000 clusters which will be found in rcs-2 , so the survey data themselves must be used to produce a proxy for cluster mass . significant , representative samples of clusters from rcs-1 have been followed up using a variety of mass estimators such as dynamical @xcite , x - ray @xcite , strong and weak - lensing ( from an acs snapshot programme , pi : loh ; acs sne cosmology project pi : perlmutter ) and sz observations . in this way , the relationship between our mass proxy ( optical richness from the survey data ) and cluster mass can be understood . one of the primary goals of rcs-1 was to place constraints on cosmological parameters ( @xmath14 , @xmath15 , @xcite ) via the growth of the cluster mass function . this demonstrated for the first time the feasibility of such an approach using an optically - selected cluster sample . this approach used the measured relation between mass and richness , but also showed that meaningful constraints could be obtained using a self - calibration technique @xcite to estimate the form of this relation from the survey data themselves . these authors demonstrate that the best constraints are obtained when accurate mass estimates are available for a subsample of clusters within the survey . it is worth emphasising that even if there is significant scatter in the relation between mass and the proxy ( as we have found for optical richness ) , it is only important that the size of the scatter be well understood . with an order - of - magnitude larger survey than rcs-1 , it becomes feasible to also constrain the equation of state of dark energy , @xmath16 , @xcite and this is in part the motivation for rcs-2 . rcs-1 also produced a significant sample of strongly gravitationally lensed arcs around the clusters found . the number and redshift distribution of these lensing clusters were used to argue about the physical properties of the clusters responsible for their lensing cross - section and the relevance of such systems to constraining cosmology @xcite . the identification of such high surface brightness , strongly - lensed galaxies is another primary science driver for rcs-2 . the massive clusters can be used as gravitational telescopes for studying high - redshift galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ) which would otherwise be too faint to observe in detail . the giant arcs can also be used as probes of the properties of the cluster lenses themselves . with a statistical sample of galaxy clusters , such as in rcs-1 , it is possible to study the properties of their member galaxies ( e.g. , their luminosity functions and blue fractions ) by stacking subsamples built from the survey data themselves @xcite . with the order - of - magnitude larger rcs-2 , it becomes feasible to measure much weaker trends and push measurements of cluster galaxies down to much lower overdensities . the addition of photometric redshifts ( e.g. , @xcite ) will allow these techniques to be extended to the field environment . such galaxy evolution studies will be explored in future work with rcs-2 . the outline for this paper is as follows . in 2 we give an overview of the survey design and observations ; 3 & 4 deal with ccd pre - processing , reduction , object detection and photometry ; 5 describes the photometric calibration via accurate fits to the star colors in our survey fields ; 6 outlines the procedure for and accuracy of the astrometric calibration . 7 & 8 describe the incorporation of additional data into our primary rcs-2 catalogs : @xmath17-band data which covers a large ( @xmath070% ) subsample of the primary @xmath18 , @xmath6 , @xmath19 survey area ; and public imaging data from the cfhtls - wide survey , respectively . 9 describes the final cleaning of the photometric catalogs : stitching into contiguous patches , removing duplicate data between overlapping pointings , and masking of artefacts . 10 summarises and describes ongoing and future work for the survey . the wide - field imaging capability of the square degree imager , megacam @xcite , on the 3.6-m cfht makes it feasible to carry out a survey covering a significant fraction of the sky in a modest amount of observing time . coupled with the exquisite seeing conditions attainable from the summit of mauna kea , such a survey can achieve impressive depths / resolution within this time . the main science drivers for rcs-2 dictate a survey area of @xmath01000 deg@xmath20 with the ability to detect galaxy clusters out to z@xmath01 using the method of @xcite . following the strategy of rcs-1 @xcite , which used @xmath11- and @xmath19-band imaging , @xmath21- and @xmath12-band filters are chosen to cover the bulk of this redshift range . in order to better distinguish lower - redshift clusters , @xmath18-band is added to accurately measure the colors of galaxies at [email protected] , once the 4000 break begins to move blueward of the @xmath6-filter . the @xmath18-band filter also helps to identify strongly gravitationally - lensed arcs around clusters , since the former tend to be relatively blue . thus , the survey comprises three - color , @xmath18 , @xmath6 , @xmath19 , imaging over the whole area . @xmath17-band imaging is obtained for the majority of this area via a data - exchange with the canada - france high - z quasar survey @xcite . megacam comprises 36 ccds arranged in a four row @xmath22 nine column grid at the prime focus of cfht . each chip is 2048@xmath224612 pixels with a pixel scale of 0.187 , allowing it to adequately sample the exquisite seeing possible from mauna kea . the spacing between ccds is approximately 13 , with larger gaps ( @xmath580 ) between the uppermost and lowermost rows and the other ccds . this gives sky coverage of [email protected] deg@xmath20 ( see fig . [ fig : ccds ] ) . observations were carried out in queue - scheduled mode on cfht on runs between semesters 2003a and 2007b inclusive . pi imaging time was granted through requests to canadian and taiwanese agencies . exposure times were set to 4 , 8 , and 6 minutes in @xmath18 , @xmath6 , and @xmath19 respectively . in 0.65 seeing , according to the megacam exposure time calculator , these should correspond to 5-@xmath8 point source limits of @xmath2325.3 , @xmath2424.8 , and @xmath2522.5 . the depth was set in the two reddest bands ( @xmath6 and @xmath19 ) by the requirement to reach @xmath26 red - sequence cluster galaxies at z@xmath01 . part way through our survey observations , it was found that 0.65 seeing was not available on as many nights as required by our program per semester . so , a two - tier strategy for @xmath6 and @xmath19-band image quality was adopted in which only half the survey would be conducted in the 0.55 - 0.75 bracket and half would be conducted in 0.75 - 0.90 . this means that the lowest richness clusters will not be detected in the worse seeing imaging ( due to the reduced depth ) all the way out to z@xmath01 , but richer clusters will . since the former are exponentially more numerous , this should have negligible impact on cosmological constraints . @xmath18-band imaging was always performed in the better - seeing bracket in order to preserve the low surface brightness requirement for the detection of strongly - lensed arcs . our final measured depths are detailed in [ sec : magcal ] . single exposures were taken at each pointing position , i.e. , no dithering was used . this is because sufficient depth is achieved in only a single exposure of 4 - 8 minutes ( depending on filter ) , so the overhead associated with reading out the ccd ( @xmath52 minutes ) becomes a significant fraction of the integration time if this is divided into two or more exposures . the priority is to cover as large an area of sky as possible to these depths in a given amount of observing time . with this strategy the survey will contain chip gaps , but these can be dealt with ( for cluster finding , etc . ) via geometric correction factors ( discussed in [ sec : stitching ] ) . gaps in the data also appear due to the occasional failure of a ccd within the mosaic camera . ) . ] individual pointings are arranged in a grid pattern in discrete patches / fields ( discussed in the next section ) . pointings are named with the patch name , followed by a letter designating the column , beginning at a running east to west . next a digit specifies a pointing s location in the declination direction , beginning at 0 , running from south to north . thus , pointings within patch 0133 are labelled 0133a0 to 0133f5 , from the south - east to north - west corners across the patch . each pointing overlaps with its neighbors by @xmath51 arcmin , typically . see fig . [ fig : patch ] . the original observing strategy was to observe a given pointing sequentially in all three filters ( @xmath18 , @xmath6 , @xmath19 ) , but the overheads associated with filter changing made this relatively inefficient and so , after the first semester , the queue observers switched to observing several neighboring pointings in a single filter and then repeating these in the next filter . @xmath17-band imaging is typically performed many months later for a given pointing . this is potentially a consideration for projects wishing to use rcs-2 for time variability studies . when conducting a large survey , it is obviously advantageous to place fields such that they overlap with other large surveys at other wavelengths , whenever possible . the rcs-2 survey area is divided into discrete patches ( where ` patch ' refers to a contiguous set of megacam pointings ) . the patches are divided approximately into two categories : those designed to overlap with other surveys ( such as swire , @xcite ) are typically a @xmath27 grid ( i.e. , 6 degrees on a side ) ; and those designed to be ` blank ' pointings typically @xmath28 . the larger patches are designed to be sufficiently large that the clustering of galaxy clusters may be used to provide additional constraints on cosmology ( e.g. , @xcite ) . these larger ` blank ' pointings are placed in the footprint of the ukidss @xcite wherever possible , so that we will obtain nir data for our optical imaging , useful for deriving stellar masses of galaxies , for example . the patch size is reduced to @xmath29 where we target other surveys , if the field size of the survey with which we overlap is smaller than @xmath30 deg@xmath20 . we choose discrete patches distributed across the sky in order to improve observing efficiency ( by distributing fields in ra ) , and to minimise the effects of cosmic variance ( although this would only be an issue for the rarest systems , such as the most massive clusters , since our individual patches are so large - the largest patches in rcs-2 are already bigger than the entire rcs-1 ! ) . most of our fields are equatorial in order to maximize follow - up from telescopes in both hemispheres . all patches are chosen to be in regions of low galactic extinction ( see fig . [ fig : map ] ) . another consideration is the avoidance of bright stars . using the bright star catalogue @xcite , patches are shifted slightly to minimise the number of stars they contain in the range @xmath31 wherever possible . stars this bright ( and indeed somewhat fainter ) can cause large reflection artefacts ( see [ sec : masking ] ) which can render significant areas of the image unusable . stars brighter than @xmath32 are potentially hazardous to the detectors and so pointings in our grid which would overlap with such stars are not observed . a list of the patches observed is given in table [ tab : patches ] . patch 1040 covers the lockman hole , also surveyed by swire and ukidss - dex ( deep extragalactic survey ) , as well as partially covered by ukidss las ; 1613 also covers a swire ( elais - n1 ) and ukidss - dex field ; 1645 covers a swire ( elais - n2 ) field . 2329 targets a deep2 field . in addition to these targeting specific surveys , our survey fields have been used by other surveys . the wigglez survey @xcite has surveyed several of our patches ( 0047 , 0133 , 2143 , 2329 , 2338 ) with galex in order to select [email protected] lyman break galaxy candidates using optical / uv colors to generate a large spectroscopic redshift sample to measure cosmological parameters via the signature of baryon acoustic oscillations imprinted on their clustering . the canada france high - z quasar survey ( cfhqs , @xcite ) uses our @xmath19-band imaging in conjunction with their own @xmath17-band imaging to search for high redshift ( z@xmath07 ) quasars . these fields total 785 one - square degree pointings of imaging data for rcs-2 . to this we add the four patches of the publicly - available wide component of the cfhls which tallies 171 deg@xmath20 in @xmath34 , @xmath35 , @xmath21 , @xmath36 , @xmath12 . the reformatting and recalibration of these data to resemble the reduction of our pi imaging data is described in [ sec : cfhls ] . thus , overall , rcs-2 comprises 955 one - square degree pointings . lcccp3.5 cm 0047 & 00:47.4 & + 00:45 & 9@xmath229 & + 0133 & 01:33.2 & -00:10 & 6@xmath226 & + 0310 & 03:10.3 & -14:11 & 9@xmath229 & initial patch extended to better cover the wmap cold spot + 0357 & 03:57.2 & -08:48 & 6@xmath226 & + 1040 & 10:40.9 & + 57:48 & 6@xmath226 & lockman hole + 1111 & 11:12.0 & -05:52 & 9@xmath229 & + 1514 & 15:14.6 & + 08:55 & 9@xmath2210 & + 1613 & 16:13.6 & + 55:00 & 6@xmath226 & elais - n1 + 1645 & 16:45.9 & + 39:35 & 6@xmath226 & elais - n2 + 2143 & 21:41.0 & -00:07 & 10@xmath2210 & + 2329 & 23:26.1 & -01:14 & 8@xmath226 & deep2 + 2338 & 23:38.5 & -09:07 & 9@xmath229 & + w1 & 02:18.0 & -07:27 & 9@xmath228 & xmm - lss + w2 & 08:57.8 & -03:18 & 5@xmath225 & + w3 & 14:17.9 & + 54:30 & 6@xmath227 & + w4 & 22:11.4 & + 01:48 & 6@xmath226 & vvds + . standard ccd pre - processing ( bias subtraction , bad - pixel masking , flatfielding , etc . ) is performed by the elixir project at cfht . for the interested reader , an excellent description of the raw megacam data , including example images at various stages of elixir reduction , may be found at http://www.cfht.hawaii.edu/science/cfhtls-data/rawdata.html . we use the default elixir reduction to handle all these preliminary steps , except for the @xmath19-band defringing which we found could be improved upon . due to the large volume of data generated by our observations , rather than starting from raw ( non - defringed ) data and attempting to generate new master fringe frames from other @xmath19-band observations taken at similar times to our observations - band data taken within a short enough time - frame to do this anyway . ] , we investigated the possibility of starting with the elixir ( non - optimally)-defringed @xmath19-band images . it was found that running these images through defringing code developed by the cfht supernova legacy survey ( snls ) team ( kindly provided by a. conley ) resulted in improvement to the defringing for all @xmath19-band images .- band fringe pattern from data taken on an early snls observing run ( using megacam ) . it then attempts to find the amplitude of the fringe pattern which must be subtracted ( the eigenvalues ) in order to remove the pattern . ] however , for a small fraction of the @xmath19-band data ( @xmath05% of all our images ) , significant fringing is still present . the initial fringe amplitude in the raw data was at a level of @xmath015% of sky . this was typically reduced to better than 5% of sky by the initial elixir defringing and reduced to negligible levels ( @xmath100.5% of sky ) by the snls code for @xmath3395% of all our data . for the remaining @xmath05% of our @xmath19-band images , the residual fringing was at worst @xmath03% of sky . thus , the most serious impact to our photometric measurements from the residual fringing would be an additional photometric error of @xmath50.03 mag in @xmath05% of the @xmath19-band photometry added in quadrature to the intrinsic poissonian error due to photon statistics . thus , this inflates the photometric errors for very bright objects where the intrinsic error is @xmath100.03 mag ; but has negligible impact on the vast majority of our objects , which are fainter and thus have larger intrinsic errors . the one aspect of our catalog construction impacted by the presence of low - level residual fringing is object detection . the implications for this are discussed in [ sec : ppp ] . we note that our final images are identical to those generated by elixir and stored in the cfht archive ( except for the @xmath19-band which have been further de - fringed ) . however , for each image , we also generate an associated bad pixel mask ( bpm ) using the procedures outlined below , which stores information about pixels contaminated with reflection haloes from bright stars , satellite trails , and cosmic ray hits , etc . an initial estimate of the world coordinate system ( wcs ) is taken from the elixir solution written in the image headers . in many cases this solution can be offset from the real pointing position by several arcseconds , and a significant fraction of these show a non - physical instrumental solution ( such that one or more of the chips appear to be overlapping or displaced from their nominal position in the camera grid ) . so , we construct a corrected wcs by running fast object - finding , with a high threshold set to find only a subsample of the brightest objects , and pattern - match these against an astrometric reference catalog . we use a model for the global placement of the chips within the camera and solve for the locations of all objects within the pointing simultaneously . this alleviates the problem with displacement of individual chips , from the elixir solution , which would otherwise cause problems when we attempt to measure offsets between the different filters for each pointing . this quick calibration follows a similar method to that which provides our final , accurate astrometric calibration , which is described in detail in [ sec : astrom ] . for each pointing , a wcs is generated for each of the three filters , @xmath18 , @xmath6 , @xmath19 , and this is used to find the approximate offset between the different filters . object finding and photometry are performed using an updated version of picture processing package ( ppp ; @xcite ) . while the basic methods are the same as those described in @xcite and @xcite , because of the large amount of imaging data involved , a considerable number of additional algorithms have been implemented to allow the procedures to be carried out as a completely automated pipeline . in the following , we provide brief descriptions of the basic methodology and the added features that allow the automation of the process . in rcs-1 , object detection was performed on coadded @xmath37 images in order to increase the overall depth of the observations and to ensure the inclusion of the very reddest objects which may have been missed if the @xmath11-band alone ( the deeper band ) had been used . as a result of the low - level ( @xmath103% ) residual fringing in a small fraction ( @xmath105% ) of the megacam @xmath19 data , it is not possible to everywhere use co - added @xmath38 images for object detection . ( the @xmath19-band fringing means that spurious objects are more likely to be detected if they sit near the peak of a fringe , whereas some real objects will be missed if they sit in the valley of the fringe pattern . the higher noise in the sky caused by fringing also artificially increases the detection threshold , meaning that some genuine objects may be missed in clean areas . ) so , for consistency , and to be conservative , object detection is performed solely on the @xmath6-band images . our depths in @xmath6 and @xmath19 are matched ( by construction ) for z@xmath01 red - sequence cluster galaxies , and so even though the catalog is formally @xmath6-selected it is also @xmath19-limited . only for galaxies redder than z@xmath01 cluster members will the catalog begin to become incomplete . therefore , the only slight limitation is that searching for extremely red objects ( @xmath6-dropouts ) is not possible with this catalog . ppp performs object detection on each chip as described in @xcite , with additional algorithms to automatically compute the detection threshold and to block out problem regions on the @xmath6-image . if artifacts ( reflection haloes and saturated columns ) associated with bright stars contaminate the chip being considered , the measurement of the sky noise level , which is used in the estimate of the object detection level , may be biased high . furthermore , these regions will also contain many false object detections . so , a mask is applied to remove contaminated areas from consideration when estimating the sky level and the noise . to mask bright stars and their halos , the approximate wcs solution derived in 3.1 ( typically still accurate to better than 1 ) is used to send a query to vizier and download a catalog of all bright stars from the tycho 2 catalog @xcite within the pointing . this bright star list is used to flag regions in which the sky is heavily contaminated by light from a bright star , and which must not be used when estimating the sky level for object detection . at this stage , a simple empirical relation between the observed size of the star halo and the cataloged magnitude of the star is used to set the mask size ( which is set to be conservatively larger than might be needed ) . later in the pipeline , a more detailed model of the properties of the reflection haloes around bright stars is used to delimit the area for object finding and to set mask flags in the final catalog ( see [ sec : masking ] ) . the masking of saturated columns is done by identifying and connecting pixels above the saturation value starting from the positions of the bright stars . the masked areas are used neither in estimating detection threshold , nor in object finding . object detection otherwise follows the same basic procedure as described in @xcite . peaks are identified as significant enhancements measured in a 3@xmath223 tapered box more than 2.6@xmath8 above the local sky ( excluding masked regions , as described above ) . two criteria based on minimum number of connected pixels in a smoothed image and the ` sharpness ' of the candidate in the unsmoothed image are used to eliminate cosmic ray and noise spike detections . total photometry of an object is derived based on the growth curve of the object which is measured in a series of concentric circular apertures around the object , masking nearby objects as required ( see yee 1991 for details ) . the growth curve is initially measured to 8.5 for all objects . the center of the apertures is determined by an iterative procedure that is accurate to fractional pixels ( down to better than 1/10 of a pixel , depending on the signal - to - noise ratio of the detection ) . to alleviate the problem of pixelation at small radii , each pixel is subdivided into 7@xmath227 subpixels before integrating the flux within an aperture . the photometric curve of growth so produced is used to identify an optimal aperture within which to measure the total magnitude following the procedure described in @xcite . if the optimal aperture of the object is smaller than the standard aperture of 8.5 diameter , the magnitude within the optimal aperture is extrapolated to the standard aperture using corrections derived from the shape of the growth curves of bright point - source objects . the standard aperture size is chosen to include close to 100% of the light for small objects under all our seeing conditions . for brighter resolved objects , which normally would still have increasing flux at the 8.5 aperture , growth curves extended to a maximum diameter of 25 are measured and used in determining the optimal aperture to make sure that the bulk of the light is included . we note that objects with an optimal aperture smaller than 1.5 are classified as non - detections . the error on the total magnitude is then calculated as the sky noise within the photometric aperture used , which is sky noise - limited for the faint galaxies of interest here . the object positions detected from the @xmath6 image are taken as the ` master ' position list . to perform photometry on the images of the other filters , the position list has to be transformed to the pixel coordinates of these images to an accuracy of about 2 pixels , so that ppp is able to determine without ambiguity the subpixel centroids of the objects . for images that are offset from the parent @xmath6 image by less than 10 pixels , or , images whose offset can be estimated to an accuracy of better than 10 pixels from their approximate wcs solution ( 3.1 ) , a simple and very fast algorithm is used to register the position file . here , 35 to 70 brightest non - saturated objects are identified in the @xmath6 image using a quick - find algorithm and used as position reference objects . local maxima in the daughter frame within 10 pixels of the expected positions of the reference objects are located and used to derive average shifts in @xmath39 and @xmath40 . a rotation , if one exists and is sufficiently small ( @xmath411 degree ) , is also measured by comparing vector angles between pairs of objects . this transformation is performed on the position list , and photometry carried out on the daughter frame . occasionally , registration of the object positions for frames for the other filters of a pointing is not so straightforward due to a significant rotation and even a small scale change . this could occur when megacam is removed and re - installed , or the optics of the camera are changed ( as in the case when one lens was inverted ) or a filter is replaced ( as in the case of the @xmath17-band filter , see [ sec : iband ] ) . in these cases , more cpu - intensive algorithms are used to deal with the registration of the position files . the basic algorithm is a brute - force technique of finding the best @xmath42-square fit of the two sets of bright position reference objects , stepping through small increments of @xmath43 shifts , rotation , and scale . the algorithm steps through increasingly more parameters in the above order , so as to minimize computational time . ppp performs the overall registration process by starting with the simplest procedure , tests the validity of the transform ( based on whether the transformed positions from the @xmath6 frame match the actual objects on the other filter frames close enough for ppp to compute an accurate fractional pixel centroid ) , and if not , performs the next more complicated matching procedure . in most cases , it is the @xmath17-band images ( discussed in [ sec : iband ] ) , sometimes taken a year or more apart from the others , and comprising images taken through two different filters over the whole survey , that require the more time - consuming registration procedures . with growth curves for each object measured in all filters , colors are measured using a smaller aperture to improve the signal - to - noise ratio by minimizing the sky noise . the color aperture " used is either a 3 diameter aperture or the optimal aperture , whichever is smaller . the same aperture is used for each filter , taking the r - band as the reference . no correction is made for seeing differences between the bands since the seeing values are typically comparable and the aperture is already relatively large compared to the psf . the total magnitude for each of the other filters is then calculated by adding the relevant color to the total magnitude estimated in the reference filter ( @xmath6 ) . the color error is then the quadrature sum of the errors in the two color apertures used to compute the color . both color errors and total magnitude errors are propagated through to the final photometric catalogs . a 5-@xmath8 reference limiting magnitude for each chip is estimated by scaling the flux of a set of bright reference stars until they reach a signal - to - noise level of five . typically for extended sources , the 100% completeness limit is about 0.6 to 0.8 magnitudes brighter than this limit @xcite . an estimate of the seeing is made by measuring the average fwhm of the reference psf stars identified in each chip . the object catalogs typically comprise around 4000 objects per ccd chip . the values of seeing measured by ppp , one value per chip , are shown in fig . [ fig : see ] . the median values of the seeing are @xmath44 = [ 0.79\arcsec , 0.71\arcsec , 0.53\arcsec , 0.67\arcsec]$ ] , but note that the distributions are broad due to our two - tier strategy . a significant fraction of our data possesses seeing better than @xmath45 in @xmath6 . star - galaxy separation is performed by comparing the shape of the growth curve of each object to the weighted average growth curve from a set of four to eight reference psfs from nearby unsaturated stars ( see @xcite and @xcite ) chosen from a list of automatically identified reference psfs from the same chip . the list of identified reference psfs is created from all bright unsaturated objects that do not have close neighbors and are deemed ` not fuzzy ' by an algorithm that iteratively eliminates objects that are ` fuzzier ' than the average . each object is then given a classification of 0artifact / spurious object ; 1 or 2galaxy ; 3star ; 4saturated @xcite . the subdivisions of types 1 and 2 for galaxies is for historical reasons related to the development of ppp and no distinction is made in practice . an object is classified as saturated if one or more pixels is above the saturation adu value as provide in the elixir fits header of the image . based on tests using simulated images , the star - galaxy classification is robust up to the 100% completeness limit under typical seeing conditions @xcite . at magnitudes fainter than 5@xmath8 , star - galaxy classification is performed on a statistical basis using the variable classifier " procedure based on the relative sharpness ( rather than absolute sharpness ) of the objects as described in @xcite . accurate photometric calibration is one of the most important aspects of any imaging survey . for rcs-2 , given the color - selected nature of our galaxy cluster sample , obtaining the highest possible accuracy in the calibration of each _ color _ is particularly important . here we describe the technique we employ to calibrate the colors of galaxies as accurately as possible using the colors of stars in our fields . the photometric catalogs generated by ppp all assume a single photometric zeropoint for all of the observations made in each of the three filters . some of our observations were made in mildly non - photometric conditions , a condition we allowed after initial tests with our color - calibration technique demonstrated that we could achieve accurate calibration without the use of standard stars . only mildly non - photometric conditions were allowed so as to not unduly decrease the depth of the observations , but this still increases the number of usable nights available to our project , since we do not require photometric conditions . in rcs-1 , the @xmath46 color calibration of each pointing was performed by requiring the colors of bright galaxies to agree with a reference pointing . the @xmath11-band zeropoint was then set by requiring the number counts of galaxies to agree from pointing - to - pointing . in principal , the uniformity of the colors of stars or galaxies may be used . however , since we are primarily interested in extragalactic studies with rcs-2 , we find it preferable to force - fit the properties of _ stars _ to agree from pointing - to - pointing in our survey , decoupling any calibration errors from the properties of the galaxies we wish to study . for example , forcing the number counts of galaxies to agree from field to field then makes it impossible to study density variations of galaxies on scales comparable to the size of each field , since we would have , by construction , forced all these values to agree . thus in rcs-2 we will use the colors of stars to accurately set the calibration of each _ color _ , and then calibrate the value of the magnitudes in one reference band by comparison with an overlapping reference survey ( 2mass ) . we use the uniformity of the colors of galactic stars in order to set the color calibration from pointing - to - pointing in rcs-2 . this method is an extension of the idea used by @xcite to calibrate multicolor ( @xmath47 , @xmath48 ) follow - up observations of rcs-1 fields . @xcite constructed histograms of objects classified as stellar and required that their @xmath49 , @xmath50 and @xmath51 ) distributions agree from pointing - to - pointing ( holding the @xmath11-band calibration fixed ) . our method is inspired by @xcite who used the uniformity of the stellar locus in high - galactic latitude fields in sdss to assess the accuracy of their ( independent ) photometric calibration ( using standard stars ) . they showed that , for high latitude ( @xmath52 degrees ) fields , the colors of faint stars are essentially identical and are dominated by stars in the galactic halo ( and thus behind the majority of the milky way dust ) . they showed that one can define principal colors along the stellar locus , and that constructing a histogram along these principal colors results in a gaussian distribution where the width is dominated by the intrinsic width of the locus ( @xmath00.02 mag ) if the photometric errors are small . thus we combine the idea of principal stellar colors for checking the accuracy of photometry @xcite with the color histogram - matching technique @xcite to form a method for calibrating imaging surveys from the principal colors of stars . our procedure is as follows . we begin by constructing a reference sample . since our primary aim is a highly uniform color calibration across our entire survey , we design a calibration set using only megacam data to calibrate our filters internally to the native megacam system . we can then transform our ( uniformly - calibrated ) catalogs to any other ( standard ) system afterwards . to build a reference set , we select a subsample of around 20 rcs-2 pointings which overlap with sdss . both datasets are corrected for galactic reddening using the map of @xcite , since the majority of the ( @xmath52 degrees ) stars lie behind the milky way dust @xcite . ( all our fields lie at @xmath53 degrees , with the exception of the cfht legacy w2 field which is @xmath54 degrees . ) for each pointing we match individual objects with their counterparts in sdss and work out the median offset from sdss for that pointing and filter . for example , for the @xmath55 pointing in the @xmath6-band filter , the natively calibrated magnitude , @xmath56 , is related to the instrumental magnitude produced by ppp , @xmath57 , by the offset @xmath58 by @xmath59 where @xmath60 and @xmath61 and @xmath62 refer to the @xmath63 object matched between sdss and megacam respectively . now , for the @xmath6-band , the color term between megacam @xmath6 and sdss @xmath6 is negligible , so this transformation effectively transforms onto the sdss ab system . for the other filters there are color terms . adopting the above procedure for all filters ( i.e. , neglecting to specifically apply a color term ) means that our reference set is calibrated onto a native megacam system which agrees with sdss for objects with the colors of the median - color objects in sdss . we derive the color terms which may be applied to convert the rcs-2 native system to sdss ab at the end of this section . we can now take each ( uncalibrated ) science pointing in turn and calibrate it to agree with the above set . we do this for every pointing , including those which were used in constructing the reference sample , to ensure complete consistency across the survey . for calibrating @xmath18 , @xmath6 , @xmath19 data we will calibrate the @xmath64 and @xmath65 colors , initially holding the @xmath6-band magnitude fixed . we start with the @xmath64 color since the red branch of the stellar locus is almost independent of @xmath65 color ( fig . [ fig : starcols ] , see also fig . 2 of @xcite ) . the @xmath64 principal color is obtained by simply subtracting the @xmath64 color of the peak of the @xmath64 historgam , which we denote @xmath66 , where @xmath67 . the resulting histogram , now centred on zero color , is shown as the upper right panel in fig . [ fig : starcols ] . the principal color used to calibrate @xmath65 is defined by fitting a linear relation to the blue part of the color color diagram ( fig . [ fig : starcols ] ) : @xmath68 \cap ( g - r)<1.158 . \label{eqn : p1}\ ] ] the @xmath64 color cut ensures that only the blue branch of the stellar locus is used . this results in a very clean gaussian in the histogram , shown as the lower right panel in fig . [ fig : starcols ] . with the above reference set in - hand , each science pointing may be examined in turn and its star colors forced to agree with that of our reference set . all objects classified as unsaturated stars with magnitude errors @xmath69 mag are extracted and color histograms built in the same way as for the reference sample . the @xmath70 ) histogram in the science pointing is cross - correlated with the of the reference pointing , and the offset @xmath71 found . this procedure is typically found to be accurate to a few hundredths of a magnitude ( @xmath72 ) , by comparing the result with an independent calibration such as sdss . in order to improve this accuracy further , a gaussian profile is fitted iteratively around the position of the peak of the histogram , and the peak position and width recorded . the offset between the gaussian - fitted peak and zero is a refined estimate of the color offset , @xmath73 . the width of the fitted gaussian , @xmath74 , is used to check for problems with the photometry / calibration method . the width of the distribution is set by the intrinsic width of the stellar locus convolved with our photometric errors , and thus should be approximately constant from pointing - to - pointing . if @xmath75 mag then the pointing is flagged as having potential problems with the photometry . the inflated width of the gaussian usually means that the @xmath64 colors are smeared out due to the the @xmath18 and @xmath6 frames not having been registered correctly ( occasionally due to the wrong pointing position being observed by the queue observers ) . a similar warning flag can be generated by flagging unusually low amplitudes of the gaussian peak ( since the number of stars in each pointing is approximately constant ) . such calibration problems are later inspected manually . the calibration in @xmath64 is thus @xmath76 where @xmath77 is the offset determined by cross - correlation of the color histograms , and @xmath73 is the offset from the peak of the iteratively - fitted gaussian . in the case of a gaussian width @xmath75 mag , @xmath73 is set to zero and a warning flag set . in a small minority of cases , this warning flag is caused by some problem other than bad registration of the images ( such as contamination by satellite trails ) . in this case , if visual inspection of the calibration histograms reveals no obvious problem , then the calibration is accepted using just the cross - correlation offset , @xmath71 . a similar procedure is applied in @xmath65 , but using the principal color p1 ( eqn . [ eqn : p1 ] ) when constructing the color histogram . note that the @xmath64 color cut used with p1 to select the blue branch of the stars gives a very sharp , clean histogram in @xmath65 . this can be done accurately since the @xmath64 color has been calibrated in the previous step . again , a cross - correlation offset followed by a gaussian fit to the located peak is performed and the same check , @xmath78 mag , made of the results . the above procedure results in very accurate calibration of the @xmath64 and @xmath65 colors ( @xmath100.03 mag ) . however , the magnitude in any individual filter has not yet been accurately calibrated and may be considerably in error by several tenths of a magnitude ( the data may have been observed through thin cloud , and even if not , we have made no explicit adjustment for atmospheric extinction ) . operationally we held the @xmath6-band fixed above , but in order to calibrate the individual magnitudes it is sufficient to calibrate any one of the individual filters , provided we hold the colors fixed to their calibrated values . to obtain a calibration , we need a survey which overlaps with objects in our sample and possesses well - calibrated photometry . sdss would be ideal , but it does not overlap with all of our pointings ( or even all of our patches ) . so , we use nir data from 2mass . @xmath79-band total magnitudes from the 2mass point source catalog are downloaded for objects in the rcs-2 footprint . a reference set is created again using the sdss - calibrated pointings , as above . thus , we can construct color color diagrams expected for stars in any combination of our ( @xmath802mass ) observed filters . we use @xmath81 versus @xmath64 as shown in fig . [ fig : gj ] . as mentioned above , we only need to calibrate one filter . although @xmath19-band is closer in wavelength to observed @xmath79-band , it does not matter which combination is used , since the fit is performed in color color space and the locus is just as well defined in these colors . is that this is the last of the three filters on which ppp performs photometry , so using it as the calibration filter is useful for catching problems with the photometry if the last filter ( @xmath18 ) fails , then all the photometry fails for that pointing . ] for @xmath81 the principal color used is : @xmath82 \cap ( g - r)\le1.2 . \label{eqn : gj}\ ] ] the offsets , calculated as above , result in a @xmath18-band shift , @xmath83 , which must be applied to each filter in such a way as to preserve the calibrated colors ; i.e. , if we re - write the original color shift ( eqn . [ eqn : dgr ] ) as shifts to the individual filter , defining the @xmath6-band shift as zero , @xmath84 and @xmath85 . the calibration offset measured from 2mass , @xmath86 , gives resulting overall corrections to the instrumental magnitudes of @xmath87 and @xmath88 to preserve the calibrated @xmath64 color , and where @xmath89 was above defined as zero . similarly , for @xmath19 the offset would be @xmath90 since its calibration was performed in @xmath65 and @xmath6 has just had @xmath86 added to it . just to reiterate the procedure , we take data , which may have been acquired in non - photometric conditions , and obtain an accurate photometric calibration simply by forcing the colors of stars to agree with each other from pointing - to - pointing ( which ensures highly uniform color calibration ) and obtain a magnitude calibration by forcing these colors to agree with a reference set including @xmath79-band magnitudes from 2mass in one of the colors . this procedure negates any need for separate correction of observational effects such as exposure time differences and atmospheric extinction . we note that the idea of using stars to provide or test a photometric calibration is not a new one ( indeed , @xcite recently presented an implementation of a similar technique ) , but we believe we are the first to implement such a method as the sole means of calibrating a survey of this size . it is worth emphasising that since the stars used in the stellar locus - fitting are behind the majority of the galactic dust , and we have reddening - corrected our reference dataset , the magnitude offsets measured between instrumental magnitudes and the reference set will correct the data to a system already corrected for galactic extinction . we note that in practice we prefer to both reddening correct our instrumental science catalogs and our reference dataset before performing the calibration . thus , the magnitude differences , @xmath91 ( e.g. , eqn . [ eqn : dgr ] ) , can be directly applied to the initial instrumental magnitude catalogs ( i.e. , not corrected for galactic extinction ) and this corrects the catalog initially to observed rather than reddening - corrected magnitudes . we find this more useful for considering observational effects ( such as checking image depth ) . we store sub - pixel interpolated versions of the @xcite dust maps ( to prevent possible sharp discontinuities across our patches caused by the large pixel size of the former ) matched to our survey regions as part of this process . these then allow straightforward reddening correction of the observed magnitudes afterwards . a second useful note about our procedure is that it is the almost vertical nature of the red branch of the stellar locus in @xmath92 which greatly simplifies the initial calibration . the final depths of our imaging data are shown in fig . [ fig : lims ] . limits refer to the 5@xmath8 point source limiting magnitude as measured by ppp . in order to assess the accuracy of rcs-2 photometry , the significant overlap with the large , well - calibrated sdss is used . in addition , this allows calculation of the transformation between the rcs2 megacam native system and the widely used sdss system . the first step is to perform object - by - object matching for each pointing , as was done to construct the reference stellar locus dataset , as described in the previous section . to elucidate this , we show the comparison of a randomly selected rcs-2 pointing in fig . [ fig : dr_one ] . this shows the comparison between @xmath6-band magnitudes in rcs-2 and sdss for this pointing . the average offset between the two sets of calibrated magnitudes is estimated via a biweight fit to the difference which is shown by the solid horizontal line . for this example pointing , the difference is 0.048 mag . the scatter is dominated by sdss photometric errors at the faint end , due to the shallower sdss photometry . at the bright end , the scatter is likely dominated by differences in the way magnitudes are measured by each survey . as described above , the rcs-2 magnitudes from ppp are essentially estimates of the total magnitude , and these are compared with sdss modelmag magnitudes . comparing colors relies on a more stable measurement than trying to estimate total magnitudes , so we begin here . [ fig : coldiff ] shows the differences in @xmath64 and @xmath65 and @xmath93-band data are described in [ sec : iband ] . ] colors measured in rcs-2 compared with sdss . @xmath18-band magnitudes have been corrected for the color term between megacam and sdss @xmath18 , as will be described below . each panel shows a different rcs-2 patch and each value in the histogram represents the biweight average offset ( as discussed above ) difference between the two surveys for a single pointing . only pointings with a significant number of objects ( typically @xmath941000 ) in common are considered . for each patch , the histogram in a given color is relatively narrow ( @xmath100.05 mag ) , implying that the color calibration produces very uniform colors . there appear to be some small systematic offsets with respect to the sdss photometry in the sense that the @xmath64 colors are systematically marginally redder than sdss ( @xmath00.03 mag ) , whereas the @xmath93 and @xmath65 colors are slightly bluer ( @xmath00.04 mag ) . there also appears to be some evidence that the size of this small offset varies from patch to patch . this might be due to inaccuracies in the assumption that the milky way stellar populations used in the color calibration are identical for every patch ( either due to intrinsic differences due to , e.g. , metallicity of the stars , or inaccuracies in the assumption that they may all be considered behind all the milky way dust , as measured by the @xcite dust map ) . nevertheless , the calibration accuracy of the colors still represents a significant improvement over that usually achievable with the typical number of standard star observations made throughout an observing run . this is even more remarkable considering many of our data were taken in slightly non - photometric conditions . [ fig : rdiff ] is constructed in the same way as fig . [ fig : coldiff ] , but the histograms now show the difference in @xmath6-band magnitude between rcs-2 and sdss . the scatter is somewhat larger than the error in the color calibration , more like @xmath00.05 mag rms , with some larger outliers . this likely reflects the limitation of using 2mass photometry to set the magnitude zeropoint . it is difficult to find a sufficient number of common stars between 2mass and typical 4-m telescope photometry , such that the stars are unsaturated in the larger telescope and sufficiently bright in 2mass to yield a high s / n measurement of the star s magnitude . even with the one square degree field of megacam , only @xmath530 stars are typically suitable for use in the magnitude calibration . differences in the shape of the response of the megacam system ( filter@xmath80ccd@xmath80optics@xmath80atmosphere ) versus sdss can be readily calculated following an extension of the above comparison . individually - matched objects are again considered , but this time the average offset in each filter for each pointing is first subtracted ( i.e. , this would make the histograms in fig . [ fig : coldiff ] and [ fig : rdiff ] all zero - valued ) . we do this because we are only interested in the _ shape _ of the response , i.e. , the slope of the magnitude difference as a function of megacam color , not its absolute value . [ fig : gcolterm ] shows this difference as a function of megacam @xmath64 color . all common , unsaturated objects ( @xmath9440 000 ) in a single patch are used . as can clearly be seen , the @xmath18-band magnitude difference between the two surveys is a strong function of @xmath64 color . the best fit relation , shown as the solid line , is given by @xmath95 as can be seen from fig . [ fig : gcolterm ] , the best - fit line passes through zero offset from sdss at the position of the median color [ @xmath96 , by construction , from eqn . [ eqn : mediansdss ] . due to the way in which we have internally color - calibrated rcs-2 , this transformation to the sdss system is now a trivial last step of the calibration procedure , which may be applied by using eqn . [ eqn : gcolterm ] on the photometric catalog , if desired . tests for color terms in @xmath6 and @xmath19 are performed in the same way . the results are shown in fig . [ fig : rzcolterms ] . the @xmath19-band shows a tiny residual color dependence @xmath97 which would make @xmath980.01 mag difference to the @xmath19-band data over the typical color range of objects and so may safely be neglected when comparing with sdss photometry . the @xmath6-band shows no measurable color term . to summarise the systematic errors of our photometric catalogues , our color calibration technique is able to produce colors calibrated to to an absolute accuracy of @xmath99 mag . this is performed independently of the magnitude calibration ( single filter ) which is somewhat less accurate ( @xmath100 mag ) due to the need to tie to the 2mass system , which has limited numbers of objects in each of our pointings . the 5-@xmath8 point source depths of our catalogs ( at which the magnitude error is @xmath50.2 mag ) are given in fig . [ fig : lims ] . our random magnitude errors are then limited by photon statistics in the usual way . one small exception to this is due to a small fraction ( @xmath105% ) of our @xmath19-band data which possesses an additional error due to residual fringing at a level of @xmath100.03 mag . this makes a negligible additional contribution to the photometric errors for the vast majority of sources of interest to us for cluster detection ( which are predominantly faint and have larger intrinsic photometric errors than this already ) . as one final test of the internal accuracy of our catalogs , we present a comparison of photometric differences between objects measured in the overlapping regions of neighboring pointings . [ fig : olaps ] shows the results for one representative patch , 2143 . for each pointing , the difference between magnitudes in each of the @xmath18 , @xmath6 , and @xmath19 filters measured within that pointing and one of its neighbors is recorded . only pairs of pointings with at least 100 objects in common are considered . the median difference between this pair of neighbors is then calculated and appears as a single entry in the histogram in the left panel of fig . [ fig : olaps ] . the median absolute deviation is recorded for the same set of objects and is entered into the histogram in the right panel . the results are plotted for every unique pair of overlaps . the _ rms _ width of the difference histogram ( left panel ) is better than 0.04 mag , in good agreement with the estimates from the sdss comparison . similarly , the average intrinsic scatter within the overlap regions agrees with this . tails toward greater values of the magnitude difference are seen , and illustrate one of the drawbacks of this comparison . the overlapping regions must necessarily come from the edges of the camera , where corrections for flatfielding , etc . are generally worst . thus , comparing the edge in one pointing with a different edge of the camera in a neighboring pointing generally produces the worst agreement , and the internal accuracy of our catalogs are likely better than this over most of the field . nevertheless , this is a useful , if somewhat pessimistic , estimate of the uncertainty . similar results are seen for the other patches . astrometry is performed in a manner similar to that used in @xcite . an instrumental solution is first generated using an observation of an astrometric standard field , taken as part of the megacam calibration procedure . this is used to map individual chip coordinates to a global instrumental solution in terms of @xmath101ra , @xmath101dec from the pointing centre . this is achieved semi - interactively using custom - written routines to fit a 2d cubic function using the idl s polywarp routine . in the absence of differential atmospheric refraction and other real - world limitations such as the reproducibility of mounting the camera on the telescope , this solution would be sufficient to describe the relative positions in every observation . in practice , it is necessary to fit for not only relative shift ( of the pointing centre relative to the expected telescope pointing position ) and rotation , but also for the differential displacement of objects across the field . this is achieved using lower - order corrections to the instrumental polynomial map , again using idl s polywarp . for each pointing , x , y positions are taken from the ppp catalog . now that the photometric calibration has been performed , it is a simple matter to extract a magnitude - selected sample in a similar way to an external astrometric reference catalog . the nominal object selection is @xmath102 , which is matched against the full depth of the usno - b1 astrometric catalog @xcite using the instrumental solution as an initial estimate of the positions . shifts , rotation and distortion are found iteratively by proceeding in small steps . after each iteration , the _ rms _ offset from the usno - b1 is measured and the results recorded after the final iteration . a comparison of object positions from an entire rcs-2 patch with positions determined from sdss is shown in fig . [ fig : astrom ] . a generous matching radius of 3 is used between the two catalogs to ensure that the difference is not underestimated due to distant matching objects being missed . the distribution of differences between object centres is shown with a gaussian of mean , @xmath103and width , @xmath104 . the distribution has a higher mean offset than might be expected given its width ( the mean and variance should be the same if the distribution is approximately poissonian ) , and so the bulk of the offset may be due to systematic differences between the sdss system and ours . a comparison between independent observations of the same objects in overlapping regions at the edges of our pointings shows that the rms difference is closer to 0.1 . regardless , our astrometric precision is @xmath105 , and this is comparable to ( or better than ) the accuracy of the photographic plates used in making the usno - b1 . this accuracy is confirmed by a large amount of spectroscopic follow - up with the wide - field spectrograph aaomega by the wigglez collaboration @xcite , and by our own follow - up observations . although the rcs-2 survey proper comprises @xmath18 , @xmath6 , @xmath19 imaging , the majority of fields now also have @xmath17-band imaging obtained for the cfhqs @xcite by reobserving our fields in @xmath17 . we now discuss how these data have been incorporated , where available . the basic procedure is similar to that of the other filters , registering each @xmath17-band image to the completed @xmath107 catalogs using the alignment method described in [ sec : ali ] . once registered , photometry could proceed through ppp as before . due to the breakage of the original megacam @xmath17-band filter ( i.mp9701 ) in june 2007 , the survey contains @xmath17-band imaging from two different filters . the original filter was replaced with a new one ( i.mp9702 ) possessing a much bluer response ( close to the sdss response ) . as a result , a significant color term exists between the two filters . for simplicity in the photometric catalogs , we decided to convert the i.mp9701 native photometry to that of the i.mp9702 native response . this has the advantage that the latter has a negligible color term compared with sdss @xmath17 , and thus our @xmath17-band data will be automatically calibrated to the ab system . the transformation between the two filters is found by comparing each with overlapping sdss photometry using pointings in common , as described for the other filters in [ sec : sdssphot ] . this is shown in fig . [ fig : icolterm ] . the conversion is @xmath108 where @xmath109 and @xmath110 refer to the i.mp9701 and i.mp9702 filters respectively . the rms scatter in this transformation is 0.034 mag . colour calibration of individual science pointings is achieved in the same way as for the @xmath19-band . stars with magnitude errors @xmath980.2 mag are selected from each pointing , and the stellar locus fitted to a reference locus in @xmath64@xmath93 , constructed as for the @xmath64@xmath65 case . the principal color used ( plotted in fig . [ fig : gr_ri ] ) is @xmath111 \cap ( g - r)<1.158 \label{eqn : p_i}\ ] ] since the other bands have already been calibrated , a single offset to the @xmath17-band is all that is required . the cfht legacy survey is a joint project between canada and france to image a number of fields using megacam . the ` wide ' component of the survey , considered here , covers 171 square degrees in @xmath112 . the observing strategy is somewhat different from rcs-2 , since multiple exposures with dithering are used at each pointing position . catalogues available to the canadian and french communities ( and later to the world ) are produced by terapix ( traitement elementaire , reduction et analyse des pixels de megacam ) . we use the t0006 internal release ( 2009 november ) of the data . briefly , the terapix reduction involves similar pre - processing to the elixir products , with the additional steps of determining astrometric solutions and combining the dithered images in each filter in each pointing into a stacked image . then object detection and photometry is performed using sextractor @xcite . in order to make the terapix catalogs resemble our method as closely as possible , we extract total magnitudes in the @xmath6-band using sextractor s mag_auto estimate . we then construct total magnitudes in the @xmath18 , @xmath17 and @xmath19 bands by constructing the relevant color with respect to @xmath6-band using aperture magnitudes measured in 4.4diameter apertures and add these to the @xmath6-band total magnitude to mimic the method used by ppp . i.e. , @xmath113 for the @xmath18-band , where the subscripts @xmath114 and @xmath115 denote total and aperture magnitudes , measured as described , respectively . star / galaxy classification is performed using sextractor s class_star parameter . terapix flag all objects brighter than @xmath116 as saturated regardless of whether they are actually saturated or not . to recover bright galaxies , which are not saturated , we follow the method of @xcite considering the half - light radius estimated by sextractor as a function of @xmath117 . the most important step in ensuring that the cfhtls data and rcs-2 are on a uniform system is to ensure that the color calibrations are the same . @xmath17-band data are first converted from the old to the new filter system using the color term described in [ sec : ical ] . the above procedures produce catalogs in which each pointing is independently calibrated in terms of photometry and astrometry . the final steps are to clean the catalogs , removing artefacts such as reflection halos of bright stars , satellite and meteor trails , etc . , and to stitch the individual pointings in a patch into a contiguous area , removing duplicate objects in the overlaps between pointings ( as described in [ sec : stitching ] ) . the @xmath18 , @xmath6 , @xmath17 , @xmath19 data are then run through the rcs-2 photometric calibration pipeline , as described in [ sec : photcal ] , [ sec : ical ] . since the terapix catalog construction already contains a photometric calibration , this typically results in small @xmath118 mag shifts in the zeropoint of each filter in each pointing . terapix provide two alternate photometric calibrations for their catalogs . the default involves calibrating a single pointing in each of their patches and then using this as a reference . the calibration of all other pointings in the patch is then determined via the overlaps of all overlapping pointings , holding the single reference fixed . terapix also provide photometric offsets which have been determined via stellar locus regression using a technique similar , but not identical , to ours . in order to ensure complete uniformity across rcs-2 , we use our stellar locus fitting and 2mass comparison to calibrate the colors and magnitudes respectively , completely independently of the elixir calibration . we find from comparison of each calibration with individual objects in sdss ( as in [ sec : sdssphot ] ) that our color calibration is somewhat more accurate than the terapix color calibration ( @xmath50.02 mag vs @xmath50.05 mag ) but that our magnitude calibration is marginally worse ( @xmath50.05 mag vs @xmath50.03 mag ) , again likely limited by the number of stars available from the overlap with 2mass and their associated photometric errors ( see lu et al . , in prep for a detailed comparison ) . it is possible that we will modify the final rcs-2 magnitude calibration to use internal overlaps as a final correction , but the current calibration is certainly sufficiently precise for our purposes . two distinct sets of artefacts are present in the data which must be masked out to produce a clean photometric catalog . the first set is predictable artefacts such as those associated with bright stars . the positions of bright stars in megacam pointings are known a priori ( once the astrometric calibration has been performed ) , and so the locations of halo reflection artefacts may be automatically predicted and masked . secondary reflections within the camera lenses , or returned reflections off the detector or camera dewar window and then off an air - glass interface in the camera result in extended reflection haloes visible around bright stars . careful examination of the images revealed three dominant reflection haloes . each halo is of similar total intensity , and so the largest of these is visible above sky noise for only the brightest stars ( typically at most a few stars per pointing ) and the smallest is visible around relatively fainter stars ( typically up to a few stars per mosaic chip ) . the outer boundaries of these artefacts are approximately circular , but the center of the reflection haloes of each type is a function of position within the camera reference frame . the halo sizes , and positions as a function of camera coordinates , were determined empirically from a set of test observations in each band , and these data were used to construct masks around known bright stars . the masking additionally requires a threshold brightness for each reflection halo ; this was again determined empirically by examining object counts radially around bright stars , and choosing limiting magnitudes for each reflection halo that ensured that regions which are significantly compromised ( either by having too many objects detected due to structure in the reflection halo , or by having too few objects due to the increased sky noise ) are masked . objects in these masked regions are retained in the photometric catalog , but a flag is set warning that they are located in a reflection artefact and thus they may be straightforwardly rejected from subsequent analysis . the second set of artefacts are those whose locations can not be predicted prior to examining the data , such as satellite trails and meteor trails ( see fig . [ fig : patch ] ) . these are dealt with in a semi - interactive manner . since these are linear artefacts , line - finding techniques akin to the hough transform ( converting the positions of objects in the ppp catalog to polar coordinates and looking for overdensities of points in this coordinate space ) can be used to flag potential artefacts . in practice , the images are examined interactively with sites of potential artefacts flagged by this technique . at this point other artefacts which may have passed the automated masking are also flagged . these include occasional reflection haloes around stars whose as - observed magnitudes are significantly brighter than their cataloged values . also , bright and very low redshift galaxies - typically subtending tens of arcseconds or more - will tend to produce many objects due to substructure in the galaxy . these regions are also masked to mitigate any influence of such objects on future analyses of object counts or clustering . finally the most common masking added by hand at this stage is of diffraction spikes which have escaped the automated masking done by ppp ( see [ sec : ppp ] ) . typically these are either from close but resolved double stars which appear as single entries in the catalogs used to initially mask the images and which produce two sets of diffraction spikes neither of which is caught by the initial mask , or diffraction spikes from stars which fall in the mosaic chip gaps . after all additional masking is complete , a mask image is constructed for each chip , with the values of the masked pixels set to indicate the source of the masking . these images also include a record of missing filter data along image edges ( due to misalignments between images of differing filters ) with missing data in each filter flagged by a different mask value . in principle , this allows one to select subsets of data masked only for certain reasons , while including other nominally masked objects ; this may be of interest under some circumstances . up until this point , the data processing has been performed on individual pointings . due to the survey strategy , overlaps occur between neighboring pointings . the final step is to ` stitch ' the overlapping pointings into single contiguous patches , removing duplicate objects which occur in the overlapping regions . this is performed in the same manner as for rcs-1 . briefly , the midpoint of the joint area between neighboring pointings is found and then each field is truncated at the midpoint . see @xcite for details . thus , only objects in unique , non - overlapping areas are retained in the stitched photometric catalog . at the same time as this stitching procedure is being performed , catalogs containing a uniform surface density of points placed at semi - random positions ( referred to as ` random catalogs ' ) are produced for the purposes of area calculation . these contain points placed across each patch at a nominal density of 0.1 arcsec@xmath119 . where these fall outside the survey geometry , or within regions masked by artefacts , the points are removed . thus , these essentially represent a sparse ` good pixel mask ' for the survey . future analyses requiring knowledge of the area considered around a given point ( such as the area surveyed within x arcmin radius of a given galaxy cluster ) can then count the number of points in the random catalog within that region to calculate the usable area . if higher density sampling is required , additional realisations at the same sampling density can be generated to increase the resolution as needed . we have presented the methodology for the reduction and precise calibration of the second red - sequence cluster survey ( rcs-2 ) . the primary purpose of this survey is to build the largest sample of optically - selected galaxy clusters out to lookback times around half the age of the universe . results from cluster - finding ( which will be presented in a forthcoming paper ) show that the red - sequence redshifts are better than in rcs-1 , due to the better uniformity and accuracy of the photometry in rcs-2 , with typical @xmath120 . a large number of projects aimed at characterizing clusters found with our technique have been ongoing for several years . an accurate understanding of the mass richness relation will allow the cluster survey to place constraints on cosmological parameters . future papers will present not only the cosmology results , but results on galaxy evolution , both within the clusters and the field . in addition to the large number of cluster science works possible , many other studies including those concerned with the properties of the milky way , such as the stellar populations and searches for dwarf galaxies , are possible . the large and accurate dataset provided by rcs-2 make it an important wide - field imaging survey of the new generation . we are grateful to the canadian and taiwanese tacs for the generous allocations of cfht time which made this project possible . we thank alex conley for providing the snls defringing code and for useful discussions on megacam data ; ben koester for his assistance with masking bright stars and artefacts ; and all our rcs and wigglez collaborators for thoroughly testing preliminary versions of these catalogs . based on observations obtained 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 institute national des sciences de lunivers of the centre national de la recherche scientifique 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 . , o. , charlot , x. , abbon , p. , aune , s. , borgeaud , p. , carton , p. , carty , m. , da costa , j. , et al . 2003 , in presented at the society of photo - optical instrumentation engineers ( spie ) conference , vol . 4841 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . m. iye & a. f. m. moorwood , 7281	the second red - sequence cluster survey ( rcs-2 ) is a @xmath01000 square degree , multi - color imaging survey using the square - degree imager , megacam , on the canada - france - hawaii telescope ( cfht ) . it is designed to detect clusters of galaxies over the redshift range @xmath11 . the primary aim is to build a statistically complete , large ( @xmath2 ) sample of clusters , covering a sufficiently long redshift baseline to be able to place constraints on cosmological parameters via the evolution of the cluster mass function . other main science goals include building a large sample of high surface brightness , strongly gravitationally - lensed arcs associated with these clusters , and an unprecedented sample of several tens of thousands of galaxy clusters and groups , spanning a large range of halo mass , with which to study the properties and evolution of their member galaxies . this paper describes the design of the survey and the methodology for acquiring , reducing and calibrating the data for the production of high - precision photometric catalogs . we describe the method for calibrating our @xmath3 imaging data using the colors of the stellar locus and overlapping two - micron all - sky survey ( 2mass ) photometry . this yields an absolute accuracy of @xmath4 mag on any color and @xmath50.05 mag in the @xmath6-band magnitude , verified with respect to the sloan digital sky survey ( sdss ) . our astrometric calibration is accurate to @xmath7 from comparison with sdss positions . rcs-2 reaches average 5@xmath8 point source limiting magnitudes of @xmath9 $ ] , approximately 1 - 2 magnitudes deeper than the sdss . due to the queue - scheduled nature of the observations , the data are highly uniform and taken in excellent seeing , mostly [email protected] in the @xmath6-band . in addition to the main science goals just described , these data form the basis for a number of other planned and ongoing projects ( including the wigglez survey ) , making rcs-2 an important next - generation imaging survey . latexl-.36em.3ex-.15em t-.1667em.7ex-.125emx
tunneling has been one of the most characteristic features of quantum mechanics since the pioneering works by hund @xcite on tunneling in atoms and molecules , by nordheim @xcite and oppenheimer @xcite on tunneling into the continuous , and by gamow @xcite , by gurney and condon @xcite , and by born @xcite with its application to @xmath0 particle decay . it should be noticed that the outgoing wavefunction method of gamow leads to eigenfunctions with complex eigenvalues , which may be expressed as @xmath1 , where @xmath2 is proportional to the decay constant of the state . hence , gamow method is equivalent to the introduction of a non - hermitian effective hamiltonian to describe the system . although non - hermitian hamiltonians are commonly used in scattering theory @xcite , standard textbook - like tunneling between sites @xmath3 and @xmath4 is always symmetric owing to the hermitian character of the hamiltonian and to the fact that the density of states for the initial and for the final states are the same . several aspects of the tunneling theory are reviewed , for instance , in the book by razavy @xcite . various signs of asymmetric tunneling rates were observed in the last @xmath5 years , however , in systems as diversified as high-@xmath6 superconductors @xcite , non - superconducting cuprates @xcite , bose - einstein condensates trapped in optical potentials @xcite , molecular electronics @xcite , and , more recently , quantum dots interacting with leads @xcite , for instance . despite the diversity of the physical mechanisms responsible for asymmetric tunneling in any of these systems , the unifying trace is that tunneling between initial and final sites involves a coupling to an external system , characterizing an open system . hence , it is fair to say that asymmetric tunneling must be a feature present in several real systems . since the hermiticity of the hamiltonian plays a part in guaranteeing the symmetry of the tunneling , could a non - hermitian hamiltonian be capable of describing asymmetric tunneling in a simple way ? in a recent letter @xcite , we presented a non - hermitian quantum model to describe the tunneling , through a @xmath7-chiral mirror , of the light fields of a system of two resonant cavities in an attempt to understand the @xmath8 asymmetry observed earlier in the transmission of circularly polarized photons through an array of asymmetric split rings @xcite . the model consisted simply in a pair of quantum oscillators coupled by a non - reciprocal interaction , and it was recently generalized to include number non - conservation and nonlinear interactions @xcite . based on this simple model , hermitian and non - hermitian dynamics of mode entanglement between the light fields in the cavities were analyzed @xcite . we remark that the model incorporated phenomenologically the non - trivial interactions between photons and the chiral mirror without a detailed treatment of the mirror . in fact , there was no degrees of freedom corresponding to the mirror in the hamiltonian , and the effect of the mirror on the dynamics was introduced phenomenologically in the model in the non - reciprocal character of the interaction between the cavities . these works hint at the suggestion that non - hermitian quantum models are useful to describe certain classes of open systems in a simplified way . interest in non - hermitian quantum mechanics was renewed after bender and boettcher @xcite presented a class of non - hermitian hamiltonians which were symmetrical under spacetime inversion ( @xmath9-symmetry ) , and which exhibited real spectra . could @xmath9 symmetry alone be responsible @xcite for both the reality of the spectra and the unitarity of the theory ? mostafazadeh @xcite showed that @xmath9 symmetry was neither a necessary nor a sufficient condition for a hamiltonian to exhibit a real spectrum , and that exact @xmath9 symmetry was equivalent to hermiticity with respect to a suitably defined inner product . recent reviews @xcite and the references therein provide a detailed exposition of the current status of the non - hermitian version of quantum mechanics . coupled systems were introduced in an analysis of optical @xmath9-symmetric structures with a balanced gain - loss profile @xcite , and the first experimental realizations of such systems appeared shortly after that @xcite . in all these systems , non - hermiticity was caused by the reciprocal , balanced interaction between a lossy and an active element . it should be noticed , however , that these pioneer experiments were classical emulations of the quantum @xmath9-symmetric theory owing to the equivalence between the electromagnetic wave equation and the schrdinger equation for certain regimes . in this paper , we analyze kinematical and dynamical aspects of a non - hermitian model describing tunneling between two sites coupled by a non - reciprocal interaction . the kinematical part consists in discussing the representation of the system in a biorthogonal family of energy eigenvectors , owing to the non - hermiticity of the hamiltonian , and correcting for the lack of strict unitarity of the theory . the dynamical part consists in computing the time evolution of the state vector describing the system in order to determine transition probabilities . we emphasize that a novel aspect of this model is that the source of non - hermiticity is the presence of a non - reciprocal coupling between two otherwise ordinary quantum systems . an advantage of this treatment is that it allows the description of some aspects of the behavior of an open system using a formalism reminiscent of that applied to closed systems , that is , wavefunctions . while a closed system may be treated by the wavefunction or by the density operator formalisms , open systems are described by a density operator obeying a master equation . however , since the number of elements of the density matrix is the square of the number of coefficients of a wavefunction , solving the master equation for the density operator may be a difficult task . therefore , we propose to describe the dynamics of an open system in a way that may be simpler than the usual master equation method at the cost of using a non - hermitian hamiltonian . hence , the approach is a phenomenological one , in which we begin with a non - hermitian hamiltonian and work out its consequences instead of attaining it from first principles in a more rigorous , yet more complex , open systems formalisms such as the feshbach projection method @xcite , the adiabatic elimination @xcite . in this section , we describe a non - hermitian model for the tunneling of an excitation between two sites , @xmath3 and @xmath4 , and coupled by the non - reciprocal interaction @xmath10 where @xmath11 is a standard two - state ladder ( raising , @xmath12 ; lowering , @xmath13 ) operator for site @xmath14 , @xmath15 is a real coupling constant , and @xmath16 is a parameter measuring the non - reciprocality of the interaction and the non - hermiticity of the hamiltonian . since each site may be in an occupied or in an unoccupied state , it is natural to represent them by fermionic two - state operators . if @xmath17 , the hamiltonian will be a non - hermitian operator owing to the non - reciprocal character of the interaction between the two sites as illustrated in figure [ fig : figure1 ] . . ] the interaction @xmath18 is represented as @xmath19 relative to the basis formed by the vectors @xmath20 , @xmath21 , @xmath22 , and @xmath23 . the basis vectors are simultaneous eigenvectors of occupation number operators @xmath24 and @xmath25 for sites @xmath3 and @xmath4 , respectively , where @xmath26 since we are interested in single - excitation tunneling between the two sites , we will restrict ourselves to the single - excitation subspace spanned by the vectors @xmath21 and @xmath22 in order to determine the excitation exchange dynamics . in this case , the interaction is represented by @xmath27 formally , this operator is a two - state non - hermitian hamiltonian given by @xmath28 where @xmath29 and the traceless @xmath30 are both hermitian operators . this hamiltonian is similar , although not identical , to the one studied by sergi and zloshchastiev in the context of the open system dynamics of a single two - state system @xcite . in their case , however , @xmath31 , which leads to dissipation , loss of quantum coherence and probability non - conservation . two - state hamiltonians having the generic form @xmath32 where @xmath33 and @xmath34 are both hermitian operators were presented and discussed earlier in the context of the quantum brachistochrone problem @xcite , or in the context of the closely related no - signaling condition violation polemic @xcite . besides , it must be pointed out that the no - signaling condition violation polemic refers exclusively to the status of @xmath9-symmetric quantum theory as a fundamental description of nature . it is undisputable that non - hermitian models are useful even if they are only effective , phenomenological theories , as used successfully in several recent cases @xcite . the non - reciprocal character of the interaction leads to a non - symmetrical , non - hermitian matrix representing the hamiltonian . a non - hermitian matrix has two sets of eigenvectors associated with its eigenvalues @xcite . the right eigenvectors are determined by @xmath35 while the left eigenvectors are determined by @xmath36 for this simple @xmath37 matrix , it is very easy to see that the two eigenvalues of @xmath18 are @xmath38 , with @xmath39 , and that the right eigenvectors are @xmath40 while the left eigenvectors are @xmath41 with @xmath42 . these eigenvectors were normalized so that the completeness relation reads @xmath43 and the pair of the sets of right and of left eigenvectors forms a biorthogonal system , with @xmath44 for completeness , we would like to present some aspects regarding the usage of biorthogonal bases to represent quantum systems , since they seldom appear in quantum mechanics . we will begin with the representation of a state with respect to the energy biorthogonal bases . using the completeness relation , equation , a state @xmath45 may be represented by @xmath46 where the expansion coefficients are given by @xmath47 . similarly , @xmath48 is represented by @xmath49 with @xmath50 . hence , there are two sets of expansion coefficients , one for kets , another for bras . as we will see next , these coefficients are not simply the complex conjugates of each other , as they would be if the hamiltonian were a hermitian operator . regarding time evolution , we assume that schrdinger equation @xmath51 is still valid . hence , using eqs . , , and , we find that @xmath52 although the coefficients @xmath53 are similar to the familiar expansion coefficients obtained with a hermitian hamiltonian , the evolution is still non - unitary owing to the non - hermiticity of the hamiltonian operator . in order to grasp this point , we turn our attention to the schrdinger equation for @xmath54 , @xmath55 since @xmath56 , the expansion coefficients @xmath57 for @xmath54 will not have a form as simple as that of the coefficients @xmath53 for @xmath58 . in fact , we find that @xmath59 which may be solved , after a slightly tedious algebra , to give @xmath60 at this point , it should be clear that neither @xmath61 nor @xmath62 can represent the normalization of a state . in fact , even the quantity @xmath63 is not the right choice , since it does not preserve the norm of the state as the system evolves . suppose that the initial state is properly normalized , i.e. , that @xmath64 . then , equations , and show that , in this case , the normalization factor is given by @xmath65\sin(\omega t ) \biggr\ } + \\ c_\mathrm{l}^- ( 0 ) { \mathrm{e}}^{{\mathrm{i}}\omega t}\biggl\{c_\mathrm{r}^- ( 0)\cos(\omega t ) + \frac{{\mathrm{i}}}{2\beta}\biggl [ c_\mathrm{r}^+ ( 0)\bigl(1-\beta^2 \bigr ) - c_\mathrm{r}^- ( 0)\bigl(1+\beta^2\bigr ) \biggr]\sin(\omega t ) \biggr\}\end{gathered}\ ] ] which is not a constant , in general . however , we can fix this unitarity problem by assuming that the proper normalized vectors are given by @xmath66 that is , the unnormalized vector multiplied by the inverse of the normalization factor , which is consistent with the prescription presented earlier by scholtz , geyer , and hahne @xcite . an alternative method to define a unitary quantum system using a non - hermitian hamiltonian was presented by mostafazadeh @xcite , consisting in redefining the inner product of the theory with a metric operator , which , in our case , would be given by @xmath67 we believe that the simple method outlined above is sufficient for our purposes . having discussed these points , we finally turn our attention to probability amplitudes , transition probabilities , and normalization . since the vector @xmath58 is not properly normalized as time passes , we define an unnormalized probability amplitude for the system initially in the state @xmath68 to be found in the state @xmath69 at time @xmath70 as the usual @xmath71 where @xmath72 are the expansion coefficients of @xmath73 , and @xmath58 is the time evolved form of the initial state @xmath68 . however , using our prescription for normalizing a state , equations and , the normalized probability amplitude is given by @xmath74 in this formalism , normalized transition probabilities would be given by @xmath75 in order to show that tunneling is indeed asymmetrical in this system , we focus on two different initial conditions : one in which the initial excitation is localized in site @xmath3 , the other in which the initial excitation is in site @xmath4 . in the former case , the initial state is given by @xmath76 while it is @xmath77 in the latter case . expansion coefficients are easy to calculate : @xmath78 and one may verify that the initial states are properly normalized . it is also very easy to verify that the two initial states are orthogonal since @xmath79 with all these coefficients , it is possible to calculate the normalization factors . for @xmath80 , we have @xmath81 while the normalization factor for @xmath82 is slightly different , being given by @xmath83 in each case , there are two unnormalized probability amplitudes that are interesting to calculate , the amplitudes for an excitation to be or not to be exchanged between sites . when the excitation is initially at site @xmath3 , the former amplitude is @xmath84 and the latter amplitude is given by @xmath85 however , when the excitation is initially at site @xmath4 , the amplitudes are given by @xmath86 as a function of the parameters @xmath87 , and @xmath42 . ] determination of all these amplitudes allows us to calculate the normalized transition probabilities . the probability for an excitation initially at site @xmath3 to be found at site @xmath4 at time @xmath70 is @xmath88 as shown in figure [ fig : figure2 ] as a function of the two adimensional parameters @xmath89 , and @xmath90 . for ease of reference , the time intervals in which @xmath91 will be called the high @xmath92 transition probability phase , and the time intervals in which @xmath93 will be the low @xmath92 transition probability phase . in the reciprocal case ( @xmath94 ) , the system spends equal amounts of time in the low and in the high @xmath92 transition probability phases . as @xmath95 departs from unity , the non - reciprocality of the interaction manifests itself . in the @xmath96 region , the system passes more than @xmath97 of the time in the low @xmath92 transition probability phase , while for @xmath98 , the opposite happens . in all cases , transition probability is a periodic function with period @xmath99 . for the inverse process , the transition probability is given by @xmath100 and an analysis similar to the above holds . from this analysis , we may hint that , in the long run , the occupation of the two sites should be distinct , as we will show briefly . figure [ fig : figure3 ] shows the transition probability ratio @xmath101 also as a function of the adimensional parameters @xmath102 , and @xmath95 . as expected , the maxima of the transition probability ratio depart from unity as @xmath95 becomes different from unity , signalling the asymmetry in the tunneling rates in this non - hermitian model . as a function of the parameters @xmath87 , and @xmath42 . ] besides , the probability for an excitation initially at site @xmath3 still be found at site @xmath3 at time @xmath70 is @xmath103 which means that @xmath104 similarly , @xmath105 confirming that there is no loss in this model , after normalizing the states . however , dissipation may be added phenomenologically to the model using well - known techniques such as introducing a complex diagonal term in the hamiltonian matrix . besides , it must be noticed that bypassing the normalization step leads , in general , to a non - unitary dynamics , that is , @xmath106 which shows that the unnormalized description allows for excitation exchange with the environment . in this regard , we would like to point out that the density operator for this system is given , in the unnormalized description , by @xmath107 inspection of the equations [ eq : c_l(t ) ] and [ eq : c_r(t ) ] for the wavefunction coefficients @xmath108 and @xmath109 is sufficient to show that there is no damping in the quantum coherence in the dynamics of this system , even in the unnormalized case . this stands in opposition with the master equation approach to open quantum systems , in which normalization is preserved but coherence is rapidly lost as the system interacts with the environment @xcite . in spite of this , long - lived coherence is observed even in very noisy environments , during photosynthesis @xcite , for instance . since preservation of quantum coherence would be an important achievement in quantum information theory , it represents an advantage to know that an open system may slow down decoherence if its effective non - hermitian hamiltonian is similar to the one described here . directing our attention to the occupation number @xmath110 for site @xmath3 , we find that @xmath111 when the excitation is initially at site @xmath3 , for instance . in this same case , the occupation number @xmath112 for site @xmath4 is given by @xmath113 figure [ fig : figure4 ] shows the occupation numbers for the two sites plotted against the parameter @xmath87 for a typical @xmath114 case when the excitation was initially at site @xmath3 . a distinct feature of the graph is that the occupation number for site @xmath3 is , on average , lower than the occupation number for site @xmath4 . clearly , in a @xmath115 scenario , the opposite would happen . , and @xmath116 as functions of the adimensional parameter @xmath87 for @xmath117 ( @xmath118 ) . excitation was initially at site @xmath3 . ] finally , having the transition probabilities , a semiclassical rate equation for this system is given simply by @xmath119 since the excitation is not lost from the system . figure [ fig : figure5 ] shows the semiclassical occupation numbers @xmath120 for site @xmath3 , and @xmath121 for site @xmath4 plotted against the parameter @xmath87 for a typical @xmath114 case when the excitation was initially at site @xmath3 . after a short transient period , the semiclassical occupation numbers are clearly separated , fluctuating around their new equilibrium values . as expected in this @xmath114 case , the occupation of site @xmath4 is greater than the occupation of site @xmath3 . the opposite would happen in a @xmath122 case . , and @xmath121 as functions of the adimensional parameter @xmath87 for @xmath117 ( @xmath118 ) . excitation was initially at site @xmath3 . ] in conclusion , we have shown that a non - hermitian quantum model describes the excitation exchange dynamics typical of asymmetric tunneling . an effectively unitary description of this system was attained by proper normalization of the state vectors . a feature common to all non - hermitian systems is that the normalization depends on the hamiltonian , having to be chosen dynamically . however , even in the unnormalized case , quantum coherence is not exponentially destroyed . this treatment suggests a way of dealing phenomenologically with irreversible interacting quantum systems using a formalism that mimics that of a closed system , and which may be , therefore , simpler than the usual open system formalism . we acknowledge dr . f. altintas for fruitful discussions , two anonymous referees for criticisms , comments , and questions that really improved the paper , and fr . saboia de medeiros ignatian educational foundation for continuous support . s. h. pan , j. p. oneal , r. l. badzey , c. chamon , h. ding , j. r. engelbrecht , z. wang , h. eisaki , s. uchida , a. k. guptak , k .- w . ng , e. w. hudson , k. m. lang and j. c. davis , _ nature _ * 413 * , 282 ( 2001 ) .	we present a simple non - hermitian model to describe the phenomenon of asymmetric tunneling between two energy - degenerate sites coupled by a non - reciprocal interaction without dissipation . the system was described using a biorthogonal family of energy eigenvectors , the dynamics of the system was determined by the schrdinger equation , and unitarity was effectively restored by proper normalization of the state vectors . the results show that the tunneling rates are indeed asymmetrical in this model , leading to an equilibrium that displays unequal occupation of the degenerate systems even in the absence of external interactions .
recent achievements in magnetic resonance force microscopy ( mrfm ) promise a single spin detection in the near future @xcite . one may expect that single spin signal will represent a random sequence of quantum jumps . the important problem for the theory is modeling of quantum jumps in mrfm and the computation of their statistical characteristics . in this paper we propose a simple model which describes quantum jumps in mrfm single spin detection . we consider the oscillating cantilever - driven adiabatic reversals technique ( oscar ) which currently is the most promising approach for single spin detection @xcite . in oscar the cantilever tip ( ct ) with a ferromagnetic particle oscillates , causing the periodic adiabatic reversals of the effective magnetic field on spin . the spin follows the effective magnetic field causing a tiny frequency shift of the ct vibrations which is measured . in the second section we consider the schrdinger dynamics of the ct - spin system which underlines our model of quantum jumps . in the third section we describe our model and present the results of the computer simulations . in conclusion we discuss our results . is the external permanent magnetic field ; @xmath0 is the rotating _ rf _ field ; @xmath1 is the magnetic moment of the ferromagnetic particle ; @xmath2 is a spin near the surface of the sample . ] we consider a vertical cantilever with a ferromagnetic particle attached to the ct and oscillating along the @xmath3-axis which is parallel to the surface of the sample . ( see fig . the hamiltonian of the ct - spin system in the system of coordinates which rotates with a circularly polarized _ rf _ field can be written as @xmath4 here @xmath5 is the effective momentum of the ct , @xmath6 is its coordinate , the first term describes the effective energy of the ct ; the second term describes the interaction between the spin and the _ rf _ field ; the third term describes the interaction between the ct and the spin ; and the last term describes the interaction between the spin and the random magnetic field which causes a deviation of spin from the effective magnetic field when the latter passes through the transverse @xmath7-plane @xcite . all quantities in ( 1 ) are dimensionless : @xmath5 and @xmath6 are expressed in terms of the `` quantum units '' @xmath8 and @xmath9 @xmath10 the dimensionless parameters @xmath11 , @xmath12 , and @xmath13 are defined as @xmath14 @xmath15 here @xmath16 is the @xmath17-component of the `` regular magnetic field '' which includes @xmath18 and the regular dipole - magnetic field produced by the ferromagnetic particle at the location of the spin ; @xmath19 is the @xmath17-component of the random field with zero average value ; @xmath20 is the magnitude of the spin gyromagnetic ratio : @xmath21 and @xmath22 are the effective spring constant and fundamental frequency of the ct ; and @xmath23 is the dimensionless time . using the parameters presented in @xcite @xmath24 @xmath25 we can estimate all parameters in ( 1 ) except of @xmath26 : @xmath27 to simplify computer simulations we considered the function @xmath13 to be a random telegraph signal with two values @xmath28 . the time interval between two consecutive `` kicks '' of @xmath13 was taken randomly from the interval @xmath29 , with the average time interval , @xmath30 , close to the rabi period @xmath31 @xmath32 we choose the initial state of the ct to be a coherent quasiclassical state , and the average spin to be pointed along the `` regular '' effective magnetic field with the components @xmath33 . below we describe briefly the results of our computer simulations . our simulations reveal the formation of a schrdinger cat state for the ct : the probability distribution function @xmath34 splits into two peaks . similar to our previous computations @xcite one peak corresponds to the average spin pointing in the direction of the effective magnetic field , while the other one corresponds to the opposite direction of the average spin . the two peaks oscillate with slightly different periods due to the back action of the spin as expected in the oscar technique @xcite . unlike our previous papers @xcite , the appearance of two peaks is not connected with the initial deviation of the spin from the direction of the effective field . it is induced by the action of the random field @xmath13 which causes this deviation in the process of the spin - ct dynamics . it was shown in @xcite that the interaction with the environment quickly destroys quantum coherence between the two peaks : the schrdinger cat state of the ct quickly transforms into a statistical mixture of the two possible ct trajectories . thus , the random magnetic field @xmath13 causes the quantum jumps : spin may `` jump '' to the opposite direction relative to the effective magnetic field and , correspondingly , the ct may slightly change the period of its oscillations . unfortunately , our present simulations consume too much computer time to be able to reveal more than one spin jump . this approach , clearly , does allow one not to study statistical characteristics of quantum jumps . consequently we developed a simplified model which describes statistical properties of the spin jumps . @xmath35 and @xmath36 kicks ; the solid line is a fit with @xmath37 ; ( b ) enlargement of ( a ) ] we assume that every `` kick '' provided by the function @xmath13 is followed by the collapse of the wave function . before the kick , the spin points in ( or opposite to ) the direction of the effective magnetic field . after the kick there appears the finite angle @xmath38 between the new direction of the effective field and the average spin . the probability for the spin to `` accept '' the `` before - kick '' direction relative to the new effective field is @xmath39 . the probability to `` accept '' the opposite direction , i.e. the probability of a quantum jump is @xmath40 . ( it is clear that a significant probability of a quantum jump appears only when the effective field passes the transversal @xmath7-plane . in the transversal plane , the effective field has the minimum value . thus , after every kick of the random field our computer code decides the `` fate '' of the spin in accordance with the probabilities of two events : to restore the previous direction relatively to the effective field , or to experience a quantum jump . in our model the ct experiences harmonic oscillations @xmath41 where @xmath42 correspond to two ct trajectories with the spin pointing in the direction of ( or opposite to ) the corresponding effective field , and @xmath43 is the ct frequency shift . the components of the effective field are given by @xmath44 from the experimental data @xcite for the ct amplitude @xmath45 nm we obtain @xmath46 . the frequency shift , @xmath43 , can be estimated as @xcite @xmath47 note that our model contains two important simplifications : first , we assume that the wave function collapse occurs immediately after the `` kick '' of the random field . thus , we ignore the finite time when the spin - ct system is in an entangled state . second , in a real situation the deviation of the spin from the effective field is a `` quasi resonance '' process caused by the cantilever modes whose frequencies are close to the rabi frequency . in our model this deviation appears as a result of the `` kick '' of the random field . . the full line corresponds to the value @xmath48 , the squares represent @xmath49 , the crosses @xmath50 , the circles @xmath51 the dashed line corresponds to the value @xmath52 . data have been obtained by varying parameters @xmath53 and @xmath30 in the ranges : @xmath54 and @xmath55 . ] below we describe the results of our simulations . [ f2 ] demonstrates a typical probability distribution of time intervals , @xmath56 , between two consecutive quantum jumps . the probability distribution is a sequence of sharp peaks at @xmath57 with the poisson - like amplitude @xmath58 ( certainly , @xmath59 at @xmath60 . ) the sharp peak appears as the probability of the quantum jump is significant when the spin passes through the transversal plane , i.e. every half - period of the ct oscillation which is equal to @xmath61 . the average value of the time interval @xmath62 was found to be @xmath63 with an error less than 3% . the standard deviation is equal to @xmath64 with the same accuracy @xmath65 we studied the dependence of the average value @xmath62 on the parameters of our model . we have found that @xmath62 does not depend on @xmath66 or @xmath67 . ( we varied @xmath66 from @xmath68 to @xmath30 and changed @xmath43 up to one order of magnitude . ) at a fixed value of the amplitude @xmath69 the value of @xmath62 is approximately proportional to @xmath70 . [ f3 ] demonstrates this dependence . the best fit for the numerical points in fig . [ f3 ] is given by @xmath71 for @xmath72 we have @xmath73 , @xmath74 . for the six fold value @xmath75 we obtained the same value of @xmath76 and @xmath77 . if we estimate the amplitude of the random ct vibrations near the rabi frequency as 1 pm , then @xmath78 putting @xmath79 and the experimental value for @xmath31 ( 6 ) we obtain @xmath80 s for @xmath81 and @xmath82 s for @xmath83 . these values are close to the experimental characteristic times of @xmath84 s and @xmath85 s reported in @xcite . next we computed the correlation function for the ct frequency shift @xmath86 where @xmath87 , and @xmath88 indicates an average over time . for different parameters : circles @xmath89 , squares @xmath90 , triangles @xmath91 . in all cases @xmath92 . the dashed curves show the exponential approximation of the correlation function @xmath93 with @xmath94 , respectively . ] in fig.4 we show the correlation function @xmath95 for three different values of parameters as indicated in the legend . as one can see , the general behavior is well described by the exponential function ( indicated by dashed lines in fig.4 ) @xmath93 . the relation between the correlation time @xmath96 and @xmath62 was found to be @xmath97 . we developed a simple model which describes quantum jumps of spins in the oscar mrfm technique and which allowed us to compute the statistical characteristics of quantum jumps . in our model the average time interval @xmath62 between two consecutive jumps and the correlation time @xmath96 are proportional to the characteristic time of the magnetic noise , and inversely proportional to the square of the magnetic noise amplitude . using experimental parameters @xcite and a reasonable value for the ct noisy amplitude we obtained the value of @xmath62 which is close to the experimental value of the characteristic time of oscar mrfm signal . this work was supported by the department of energy ( doe ) under contract no . w-7405-eng-36 , by the defense advanced research projects agency ( darpa ) , by the national security agency ( nsa ) , and by the advanced research and development activity ( arda ) . mamin , r. budakian , b.w . chui , d. rugar , phys . lett . * 91 * , 207604 ( 2003 ) . d. mozyrsky , i. martin , d. pelekhov , p.c . hammel , appl . phys . lett . * 82 * , 1278 ( 2003 ) . berman , v.n . gorshkov , d. rugar , v.i . tsifrinovich , phys . b * 68 * , 094402 ( 2003 ) berman , f. borgonovi , g. chapline , s.a . gurvitz , p.c . hammel , d.v . pelekhov , a. suter , v.i . tsifrinovich , j. phys . a : math . gen . * 36 * , 4417 ( 2003 ) . berman , f. borgonovi , h.s . goan , s.a . gurvitz , v.i . tsifrinovich , phys . b * 67 * , 094425 ( 2003 ) berman , f. borgonovi , v.i . tsifrinovich , quant - ph/0306107 ( 2003 ) . stipe , h.j . mamin , c.s . yannoni , t.d . stowe , t.w . kenny , d. rugar , phys . rev . lett . * 87 * , 277602 ( 2001 ) . berman , d.i . kamenev , v.i . tsifrinovich , phys . a * 66 * , 023405 ( 2002 ) .	we propose a simple model which describes the statistical properties of quantum jumps in a single - spin measurement using the oscillating cantilever - driven adiabatic reversals technique in magnetic resonance force microscopy . our computer simulations based on this model predict the average time interval between two consecutive quantum jumps and the correlation time to be proportional to the characteristic time of the magnetic noise and inversely proportional to the square of the magnetic noise amplitude .
the stability problem for a schwarzschild black hole in the form of a pure metric " perturbation analysis was settled by t. regge and j. a. wheeler in a classic 1957 article @xcite . the main practical achievement of this work was undoubtedly the formulation of a gauge transformation approach that allows a complete radial / angular separation of the einstein equations in the two cases of odd and even parity , preliminarily established , reaching a formal solution in the axial ( or _ magnetic _ ) case : the so - called _ regge - wheeler equation_. due to mathematical complications , however , the full analysis was only completed thirteen years later , after the work of mathews @xcite , edelstein - vishveshwara @xcite and zerilli @xcite that either provided a more rigorous approach to the use of tensor harmonics , or resolved some compatibility problems in the analytic treatment of the system , or else provided the final form of the equation for the radial perturbation functions in the polar ( or _ electric _ ) case : the so - called _ zerilli equation_. the search for simplicity also led these authors to exploit some useful but not general relations between curvature tensors , like those derived by eisenhart @xcite , valid to the first order and/or only in the schwarzschild case ( latin tensor indices are used for consistency with the implemented algorithms ) : @xmath0 where @xmath1 is the perturbation tensor and @xmath2 are the perturbed parts of the affine connections , ricci and einstein tensors . + on this way , after having performed the gauge transformations for each parity case , we are left with two systems of , respectively , three and six independent radial ordinary differential equations in two ( namely @xmath3 ) and three ( namely @xmath4 ) arbitrary perturbation functions , to be determined . the magnetic system is straightforwardly reduced to a single first - order equation in @xmath5 which , by a simple variable substitution , leads to the final result : @xmath6q(r)=0\ ] ] ( _ regge - wheeler equation _ ) , where @xmath7 and @xmath8 ( the first called the _ tortoise coordinate _ by wheeler as a citation of the zeno paradox ) are ( implicitly or not ) defined by : @xmath9 where @xmath10 are arbitrary functions to be determined in each particular case and @xmath11 , in this schrdinger - like equation , plays the role of an effective potential . + finally , the electric system , by a more complex change of variables procedure , necessary to deal with the terms in the wave number @xmath12 , was found by zerilli @xcite to be represented by an equation formally equal to ( [ schr ] ) ( the _ zerilli equation _ ) only with a different ( but still algebraic ) expression of the potential . since all the variables of the system are mutually expressed by regular algebraic relations , the stability problem , analyzed by substituting different forms of @xmath12 into ( [ schr ] ) , can be extended in its validity to the whole perturbation . the aim of the present work is to show that , with the essential help of computer algebra software , a similar analysis can be carried out for both parity cases , dealing with the more general spacetime of a spherical , non - rotating , eventually charged collapsed object , therefore allowing specialization not only to the schwarzschild but also to the reissner - nordstrm metric . to begin a less rigid analysis than that induced by the formulation of equations ( 13 ) , we deal first of all with the full expression of the einstein tensor which , viewed as a function of the metric and its ordinary partial derivatives , reads as : + + here , to the unperturbed metric , which refers to the usual covariant expression of the line element belonging to a generic spherically - symmetric expression depending on the radial arbitrary functions @xmath13 : @xmath14 must be added the perturbation terms , whose time dependency is represented only by a @xmath15 factor while the angular one is limited to functions of the polar angle @xmath16 . + since `` * * components * * '' , mathtensor s routine for calculating curvature tensors from an input metric , has the useful feature of applying arbitrary combinations of mathematica commands to each component @xcite , the first order form of the einstein tensor can be readily obtained by appending to the metric input file , where the regge - wheeler - gauge perturbation terms are added with a small parameter " @xmath17 as a factor , a line like : .... compsimp[a_]:=simplify[normal[series[expand[a],{q,0,1}]]/.q->1 ] .... there are only three non - zero components @xmath18 of the tensor equation that replaces the ( 1 ) : @xmath19 ( where the subtracted quantity is the unperturbed metric tensor ) and that substantially reproduce , in their radial form , the solution system @xcite-(sys . 2 ) , once the angular terms coming from the perturbation , where they are represented by @xmath20 ( @xmath21 being the legendre polynomial to the multipolar order @xmath22 ) , are fully simplified through the following relations : @xmath23\\ f'(\theta)&=-l\,(l+1)\sin\theta\;p_{l}(\cos\theta)\\ f''(\theta)&=l\,(l+1)\,[l\;p_{l-1}(\cos\theta)-(l+1)\cos\theta\ ; p_{l}(\cos\theta)]\end{aligned}\ ] ] and , being completely factored , are consequently eliminated . + the first two equations ( from now on , almost everywhere in the rest of this paper , the notation @xmath24 will be adopted -not to be confused with the definition of the metric s radial function @xmath25 ! ) turn out to be of the first order with non - mixed dependence on the functions derivatives in such a way that a single second - order differential equation can be obtained by : + + analyzing the dependence on @xmath12 , it is readily seen that to impose the equation ( [ schr ] ) with the conditions ( [ varch ] ) , a procedure of polynomial coefficient identification is required , carried out in the following way : + + here we deal with the seven non - zero components of the perturbed einstein tensor ( the complementary set with respect to the previous three magnetic " components ) , whose angular parts , coming from the perturbing gauge function @xmath26 , are factored , once the substitutions @xmath27\\ f''(\theta)&=\frac{l\cos\theta}{\sin^2\theta}\;p_{l-1}(\cos\theta)-l\,(\cot^2\theta+l+1)\;p_{l}(\cos\theta)\end{aligned}\ ] ] are performed , the resulting expressions simplified and the original four radial perturbation functions are reduced to three by the identification @xmath28 . + of the six independent linear differential equations so obtained ( the diagonal @xmath16 and @xmath29 terms being equal ) , four turn out to be of the first order , unlike the schwarzschild - specialized system which instead has three first order equations , and a completely algebraic variable - elimination procedure allows the derivation of a first integral condition which is the generalization of @xcite-(eq . 10 ) . as in zerilli s procedure @xcite , more elaborate but always tied to a polynomial coefficient identification principle , this algebraic condition plus three of the previous equations can be treated , with a double function substitution and the change of the radial coordinate , to form a new system of seven differential equations in four variables ( plus the derivative of the new radial coordinate with respect to @xmath30 ) , this time non - linear but independent of @xmath12 , to which the formal definition of the effective potential must be added . three of these unknown functions are then found , after a cascade of algebraic eliminations ( which do nt show the residual arbitrariness found by zerilli in the schwarzschild treatment ) , to be quite simply dependent on the fourth , which satisfies a final second - order very complex equation , fortunately analytically solvable when specialized to the schwarzschild and reissner - nordstrm metrics . a comprehensive table of functions and variables in the three cases general static spherically - symmetric form ( g ) , schwarzschild ( s ) and reissner - nordstrm ( rn) can be sketched , the latter two obtained , respectively , by the two substitutions \{@xmath31 } and \{@xmath32 } : + * schwarzschild : @xmath33 * reissner - nordstrm : @xmath34 ^ 2\;r^4}\end{aligned}\ ] ] with @xmath35\;r^4\\ & + 12m\;[6m^2-(19\lambda-3)q^2]\;r^3 - 3q^2\;[108m^2-(38\lambda-3)q^2]\;r^2\\ & + 342mq^4r-117q^6 \end{split}\ ] ] as a final remark , it is straightforward to verify that the two different kinds of tortoise coordinates in the general case reduce to one for all metrics having @xmath36 and that as expected , @xmath37 holds for both the magnetic and the electric parities . 9 t. regge , j. a. wheeler , _ phys . _ , * 108 * , 1063 ( 1957 ) . j. mathews , _ _ , * 10 * , 768 ( 1962 ) . l. a. edelstein , c. v. vishveshwara , _ phys d _ , * 1 * , 3514 ( 1969 ) . f. j. zerilli , _ phys . _ , * 24 * , 737 ( 1970 ) . l. p. eisenhart , _ riemannian geometry _ , cap . vi , princeton university press , princeton , 1926 . e. d. fackerell , _ a p. j. _ , * 166 * , 197 ( 1971 ) . l. parker , s. m. christensen , _ mathtensor- a system for doing tensor analysis by computer _ , addison - wesley , 1994 .	an efficient approach to tensor perturbation calculations by proper use of computer algebra methods is described , reaching the sufficient generality required for a comprehensive analysis of the schwarzschild and reissner - nordstrm metric stability problem .
the reeb foliation is a codimension one smooth foliation of the @xmath3-sphere @xmath0 obtained by glueing two reeb components @xmath4 and @xmath5 . since the one - sided holonomies of the reeb components along @xmath6 and @xmath7 are trivial , the reeb foliation is not analytic ( `` haefliger s remark '' ) . on the other hand the @xmath8-jet space @xmath9 for a function of @xmath10 variables carries the canonical contact structure . it is contactomorphic to the unit sphere @xmath11 minus any point . here @xmath12 has the standard contact form @xmath13 ( @xmath14 , @xmath15 for coordinates @xmath16 of @xmath17 ) . thus we may regard a codimension-@xmath10 submanifold @xmath18 as a system of @xmath10 first - order partial differential equations ( for implicit functions ) . if @xmath19 and @xmath20 , the system is completely integrable and regular , and therefore defines a codimension one foliation @xmath21 on @xmath22 . the leaves of @xmath21 are legendrian submanifolds of @xmath12 corresponding to the solutions . in this article we construct an embedding of @xmath0 into the standard @xmath2 so that the image has the reeb foliation @xmath21 by legendrian submanifolds . this example shows that even a non - taut foliation can be a family of legendrian submanifolds of @xmath23 . moreover we prove [ main ] there exists a smooth family @xmath24 of codimension-@xmath25 submanifolds of @xmath2 such that 1 . @xmath26 is the standard @xmath27 , 2 . @xmath28 is an embedded contact submanifold for @xmath29 , 3 . @xmath30 admits a reeb foliation by injectively immersed legendrian submanifolds of @xmath2 , and 4 . @xmath31 is an embedded overtwisted contact submanifold for @xmath32 . the foliated submanifold @xmath33 is obtained by joining two great circles @xmath34 through the legendrian torus @xmath35 . the family @xmath28 is obtained as a byproduct in the process of isotoping @xmath36 to the unknot . the author is seeking the converse approach , i.e. , to find a foliated submanifold by using contact topology or open - books ( see remark 1 in 2 ) . let @xmath37 be the natural projection of @xmath2 to the @xmath25-simplex @xmath38 , which sends the legendrian @xmath25-torus @xmath39 to the barycenter @xmath40 . the set @xmath41 contains the great circles @xmath42 where @xmath43 denotes the vertex @xmath44 . except them @xmath45 is a @xmath46-fibration . on the other hand , @xmath47 is a @xmath48-fibration . now we take the coordinates @xmath49 on @xmath50 by putting @xmath51 for @xmath52 , i.e. , @xmath53 let @xmath54 be the standard @xmath55 . we deform @xmath54 with the help of a certain family of simple curves @xmath56 depicted in fig.[fig ] ( @xmath57 ) . note that @xmath58 has a break point @xmath40 while @xmath59 and @xmath60 are smooth on @xmath61 . on @xmath50 and its parametrization by @xmath62,height=132 ] we generate @xmath63 by moving the intersection of the `` wall '' @xmath64 with the fibre @xmath65 for @xmath66 . then we can see that @xmath28 realizes the join of two large circles @xmath67 and @xmath68 . now we give a precise definition of the curve @xmath69 . put @xmath70 for @xmath71 $ ] , and take a smooth function @xmath72 and a smooth odd function @xmath73 such that @xmath74 the inverse function @xmath75 of @xmath73 is defined on @xmath76 $ ] . it is smooth on @xmath61 ( @xmath77 ) . we put @xmath78 , and take the curve @xmath79 next we show that @xmath28 is a smooth submanifold . by moving the @xmath25-torus @xmath80 for @xmath81 , we see that @xmath82 is diffeomorphic to @xmath83 . moreover @xmath28 is topologically the join @xmath84 . thus it only remains for us to examine the smoothness of @xmath28 along @xmath85 . we restrict ourself to the connected component of @xmath85 corresponding to @xmath86 and omit the other component . we put @xmath87 where @xmath88 $ ] is a parameter to be eliminated . then @xmath89 is a smooth disk since it tangents to the real @xmath25-plane @xmath90 at @xmath91 . since the function @xmath73 smoothly tangents to @xmath92 at @xmath91 , @xmath28 is a smooth @xmath3-sphere . next we consider the ( non-)integrability of the restriction @xmath93 of the standard contact form @xmath94 . using @xmath95 as coordinates of @xmath96 , we can write @xmath97 here the sign of @xmath98 with respect to @xmath99 coincides with that of @xmath100 , and that of @xmath101 . more generally , if a submanifold @xmath102 is presented by a simple curve @xmath103 on @xmath104 , the nagative areal verocity @xmath105 still presents the non - integrability of @xmath106 . in the case where @xmath107 , the integrability means the vanishing of the areal verocity . that is why the curve @xmath58 is broken into two rays to / from the origin @xmath40 , and @xmath30 is non - analytic . on the other hand , for cylindrical coordinates @xmath108 , @xmath109 and @xmath110 are written as @xmath111 this implies that the sign of @xmath98 everywhere coincides with that of @xmath101 . now we show that the foliation of @xmath33 is a reeb foliation . the definition of @xmath33 is @xmath112 where @xmath113 $ ] is a parameter to be eliminated . on the open solid torus @xmath114 , we have @xmath115 thus the surface of @xmath116-revolution of the graph of @xmath117 is a leaf . similarly , we can describe the foliation on @xmath118 . these foliations spiral into @xmath119 and form a transversely oriented reeb foliation , to which the positive hopf link @xmath120 is positively transverse . finally we see from @xmath121 that the positive hopf band @xmath122 is a supporting open - book for @xmath29 . on the other hand , the negative hopf band @xmath123 on @xmath124 is a supporting open - book for @xmath125 . thus @xmath31 is overtwisted . indeed it has the half - lutz tube @xmath126 . moreover , since we can reverse the orientation of @xmath0 by a diffeotopy , we obtain the following `` negative stabilization '' lemma . this ends the proof . the overtwisted contact submanifold @xmath127 is diffeotopic to the standard @xmath128 . particularly @xmath129 is differential topologically unknotted , but contact topologically knotted . any closed oriented @xmath3-manifold admits an open - book decomposition ( alexander @xcite ) . we can associate to it a contact structure ( thurston - winkelnkemper @xcite ) as well as a spinnable foliation ( see @xcite ) . further any contact structure is supported by an open - book decomposition ( giroux @xcite ) . using this result , the author constructed a certain immersion of any contact @xmath3-manifold into @xmath130 or @xmath2 ( @xcite ) . this construction was generalized to any dimension , i.e. , @xmath131 or @xmath132 by martnez torres ( @xcite ) . the author proved that any / some contact structure of @xmath133 can be deformed into some / any spinnable foliation ( @xcite , see also @xcite ) . he also proved that a certain higher dimensional contact structure can be deformed into a foliation ( @xcite ) . it is interesting to generalize the present result to these cases . the author would like to thank the anonymous reviewer(s ) for encouragement to include intuitive descriptions and a figure , which have made the article easier to access for general readers . 00 j. alexander , a lemma on systems of knotted curves , proc . 9(1923 ) , 9395 . j. etnyre , contact structures on 3-manifolds are deformations of foliations , int . notices 14(2007 ) , 775779 . e. giroux , gometrie de contact de la dimension trois vers les dimensions suprieures , proc . icm2002 ( beijing ) vol.ii , 405414 . d. martnez torres , contact isometric embeddings in standard contact spheres via approximately holomorphic geometry , preprint ( 2010 ) . a. mori , a note on thurston - winkelnkemper s construction of contact forms on @xmath3-manifolds , osaka j. math . 39(2002 ) , 111 . a. mori , global models of contact forms , j. math . tokyo 11(2004 ) , 447454 . a. mori , reeb foliations on @xmath2 and contact 5-manifolds violating the thurston - bennequin inequality , preprint ( 2009 ) , arxiv:0906.3237 . w. thurston , h. winkelnkemper , on the existence of contact forms , proc . 52(1975 ) , 345347 .	we realize the reeb foliation of @xmath0 as a family of legendrian submanifolds of the unit @xmath1 , moreover we construct a deformation of the standard contact @xmath0 in @xmath2 , via a family of contact submanifolds , into this realization .
solar - type stars are prime targets for many studies in astrophysics . their spectra are rich in metallic lines , which allow precise determinations of the fundamental stellar parameters and elemental abundances through different well - established techniques ( e.g. , excitation and ionization equilibria or spectral synthesis ) . they are also long - lived and have a large age dispersion , probing a considerable fraction of the history of the milky way . furthermore , the compositions of their atmospheres remain almost unchanged ( with the exception of li , be and b ) throughout their evolution on the main sequence . all these properties make solar - type stars the ideal candidates to study time - dependent processes , such as the chemical evolution in the solar neighborhood , and many successful examples of this application can be found in the literature ( e.g. , @xcite ) . the addition of radial or spatial velocities for these objects provides an approach to study dynamical processes within the disk , such as radial migration and kinematical heating ( e.g. , @xcite ) . local stellar samples with both kinematical and chemical information may also be used to identify stars from different galactic components within the thin - disk dominated population in the solar neighborhood ( @xcite ) . in order to improve the accuracy of these studies and to extend them over larger volumes , massive spectroscopic surveys such as segue ( @xcite ) and rave ( @xcite ) have been developed , with additional projects expected in the near future ( e.g. , galah ; @xcite ) . although its main scientific goal is the study of the formation and evolution of giant planets , brown dwarfs ( bds ) and low - mass stars , it was soon realized that the multi - object apo radial velocity exoplanet large - area survey ( marvels ; @xcite;@xcite ) could also contribute to a better understanding of the chemical and kinematical evolution of the solar neighborhood . during its four - year operation ( 2008 - 2012 ) as part of the third phase of the sloan digital sky survey ( sdss - iii ; @xcite ) , a sample of @xmath03,300 fgk stars with 7.6 @xmath1 v @xmath1 12.0 ( many of which were never previously analyzed ) had their radial velocities ( rvs ) monitored in the search for companions . the targets were selected according to a limited number of well - defined criteria that did not explicitly include any cuts based on the metallicities , activity levels and ages of the stars ( for more details , see @xcite ) . the final sample thus presents a well characterized selection function , providing a statistically homogeneous data set . the kinematical analysis will benefit from the precise rvs delivered by the marvels survey ( @xmath2 100 m s@xmath3 ) . the chemical analysis , on the other hand , depends on the availability of precise atmospheric parameters ( effective temperature , , and surface gravity , ) and chemical abundances ( metallicity , [ fe / h ] , and @xmath4-element content , [ @xmath4/fe ] ) . stars with rv variations suggestive of the presence of companions were selected for a more detailed study , which included the acquisition of high - resolution ( r @xmath5 30,000 ) spectra to determine their fundamental parameters by applying standard spectroscopic techniques ( excitation and ionization equilibria of fe i and fe ii lines ; see the details in @xcite ) . stars without detected companions , which correspond to @xmath080% of the survey targets , were not be subjected to a similar high - resolution spectroscopic follow - up and thus the stellar characterization had to rely solely on the marvels data . although the numbers of visits for each target ( typically @xmath520 ) were able to produce final combined spectra with high signal - to - noise ratios ( s / n @xmath5 100 ) , the moderate resolution ( r @xmath0 12,000 ) and somewhat limited wavelength range ( @xmath05000 - 5700 ) of the data prevented the usage of the classical analysis mentioned above . in this resolution regime , most of the lines are blended with neighboring features and the stellar characterization through the measurement of equivalent widths ( ews ) of individual iron lines is not feasible . recent large spectroscopic surveys operating in the low - to - intermediate resolution ( @xmath02,000 20,000 ) regime have developed pipelines that rely exclusively , or at least partially , on the spectral synthesis technique in which the stellar parameters are determined through a comparison of the observed spectra with an extensive library of previously calculated synthetic ones . as examples , we can mention the pipelines from segue ( sspp ; @xcite ) , rave ( @xcite ) , lamost ( @xcite ) and ambre ( @xcite ) . although spectral fitting is a powerful technique and provides precise results for high - quality data ( e.g. , @xcite ) , it has some drawbacks : ( 1 ) dependency on a very detailed and extensive list of spectral lines ( many of which may have poorly determined atomic parameters ) ; ( 2 ) the need to accurately know the broadening parameters ( instrumental profile , macroturbulence and rotational velocities ) . ( 3 ) stronger correlations ( relative to the excitation and ionization equilibria method ) between the resulting atmospheric parameters ( see , e.g. , @xcite ) . for the specific case of marvels , the relatively small wavelength coverage ( @xmath6 700 ) combined with the moderate resolution can significantly decrease the accuracy of the spectral synthesis technique . therefore , a different method is necessary to efficiently and accurately derive the parameters of the stars that did not have high - resolution follow - up spectra . the purpose of this work is to develop and validate an alternative approach based on spectral indices specific spectral regions combining multiple absorption lines into broad , blended features to determine atmospheric parameters directly from the marvels spectra , without any other priors . indices have been successfully applied before to derive information on mean stellar ages and metallicities of populations of galaxies and stars ( e.g. , @xcite ) , as well as atmospheric parameters for target selection purposes in planet search surveys ( e.g. , @xcite ) . because a manual object - by - object analysis of a numerous sample such as that from marvels is prohibitively time - consuming , we have developed three codes to automate the determination of stellar parameters . this paper describes the proposed automatic approach to perform the stellar characterization of marvels targets and validates its results using data from the survey and the elodie stellar library ( @xcite , @xcite ; see website for the most updated 3.1 version ) . the paper is organized as follows . section [ data ] presents the data used to build the calibrations that were later adopted for the derivation of the atmospheric parameters . a sample of marvels stars utilized to test the precision of the results is also described . the definition of the spectral indices is detailed in section [ indices ] , while section [ pipeline ] is devoted to a thorough description of the method . the four steps of our analysis ( normalization of the spectra , measurements of ews , construction of the calibrations and derivation of atmospheric parameters ) are explained in separate subsections . the results obtained with the spectral indices are shown in section [ results ] , along with a discussion regarding their accuracy . finally , our concluding remarks are presented in section [ conclusions ] . two different samples of stars were used in this work . the first is formed by stars with well - known atmospheric parameters ( derived from the analysis of high - resolution spectra ) and was utilized to construct the calibrations used to obtain the atmospheric parameters ( , [ fe / h ] and ) , thus being referred to as the _ calibration sample_. the second contains a subset of marvels targets for which independent high - resolution spectra and precise atmospheric parameters are available . as its purpose was to check the performance of the spectral indices approach , we called it _ validation sample_. both samples are described in more detail below . the calibration sample is composed of 309 stars ; their high - resolution spectra , effective temperatures , surface gravities and metallicities were taken from @xcite and @xcite . figure [ samples ] shows the distribution of these stars ( open circles ) in parameter space ; it is clear that this sample has a good coverage in the following intervals : 4800 k @xmath7 @xmath7 6500 k , 3.60 @xmath7 @xmath7 4.70 and @xmath80.90 @xmath7 [ fe / h ] @xmath7 @xmath90.50 . the majority of the stars were taken from @xcite , but the 17 stars classified as giants by the authors were not included here . this cut was done because their spectra are considerably different from those of dwarfs and subgiants and a proper analysis would require a distinct set of spectral indices . the remaining 291 stars ( 262 dwarfs and 29 subgiants ) all have high - resolution ( r @xmath0 48,000 ) and high - quality ( s / n @xmath5 200 per resolution element at @xmath10 6700 ) spectra obtained with the fiber - fed extended range optical spectrograph ( feros ) spectrograph ( @xcite ) attached to the mpg / eso 2.2 m telescope ( la silla , chile ) , under the agreement between eso and observatrio nacional ( mct ) . additional details about the sample selection , observations and data reduction are given in @xcite . the authors also describe the homogeneous derivation of the atmospheric parameters ( and ) and metallicities ( [ fe / h ] ) that were utilized in this work . in summary , the standard spectroscopic analysis based on the excitation and ionization equilibria of a carefully selected list of fe i and fe ii lines was used to determine the results in an automated procedure . in order to better populate the region [ fe / h ] @xmath11 of the parameter space , the above sample was complemented with stars from @xcite . nine stars are common with @xcite and were considered only for comparative purposes ( see below ) . the remaining 18 stars have high - resolution ( r @xmath0 48,000 ) and high - quality ( s / n @xmath5 300 ) spectra acquired with the feros spectrograph fed by the eso 1.52 m telescope . the details about the sample selection , observations and data reduction for these objects can be seen in @xcite . a description of the iterative method used to determine the atmospheric parameters ( and ) and metallicities ( [ fe / h ] ) adopted here are also provided by the authors . briefly , the effective temperatures were estimated from the arithmetic mean of the values derived from photometric calibrations and h@xmath4 profile fitting . the surface gravities were derived using these effective temperatures and also stellar luminosities and masses . the luminosities were calculated from hipparcos parallaxes and v magnitudes . the masses were obtained from interpolation in grids of evolutionary tracks using the effective temperatures , luminosities and metallicities . finally , the metallicities were determined from a differential analysis relative to the sun using fe i and fe ii lines . the 291 stars from @xcite and 18 from @xcite compose the final calibration sample used in this work . although the two subsamples were analyzed with somewhat different methods , we find that the resulting parameters are consistent . for the 9 stars in common , the average differences are ( in the sense ghezzi - del peloso ) : @xmath12 = @xmath84 @xmath13 41 k , @xmath12[fe / h ] = @xmath80.04 @xmath13 0.04 dex , and @xmath12= 0.00 @xmath13 0.12 dex . in this paper , the spectra and parameters from @xcite were adopted for these nine stars . we note that these parameters are consistent with those from many other catalogs of atmospheric parameters available in the literature ( see , e.g. , table 4 of @xcite ) . to ensure consistency between the calibration and marvels survey data ( the latter is described in section [ testsample ] ) , the feros spectra were degraded and resampled to the marvels resolution and sampling . the iraf tasks ` gauss ` ( with a value of 8.0 for the _ sigma _ parameter ) and ` dispcor ` ( with a value of 0.154 for the _ dw _ parameter ) of were used , respectively , to accomplish this . the spectra were also trimmed with iraf s task ` splot ` in order to keep only the region 5100 - 5590 , which is present in the majority of the marvels spectra . the validation sample consists of the full set of 30 stars that currently have both low resolution marvels spectra and precise atmospheric parameters resulting from the analysis of high - resolution spectra . sixteen of these stars are marvels targets with detected companions which were or are currently being subjected to more detailed analyzes . their high - resolution spectra were obtained with the arc echelle spectrograph ( arces ; @xcite ) attached to the astrophysical research consortium ( arc ) 3.5 m telescope at the apache point observatory ( apo ; new mexico , usa ) and/or the feros spectrograph attached to the mpg / eso 2.2 m telescope . the resolutions are @xmath031,500 and @xmath048,000 , respectively , and all spectra have s / n @xmath5 100 per resolution element . their atmospheric parameters were derived following the method described in @xcite and are given in table [ marvels_cand_param ] . briefly , they were derived from two independent pipelines , both based on the technique of excitation and ionization equilibria of fe i and fe ii lines . given the consistency of the results , the two sets of parameters were combined through a weighted average , using the internal uncertainties from each pipeline as the weights . the other 14 stars are known planet - hosts which were used in the marvels survey as reference objects for the rv determinations . their atmospheric parameters are the arithmetic average of the values taken from multiple sources in the literature ( see table [ marvels_ref_param ] ) . in this table , the uncertainties correspond to the standard deviations of the average values from the literature ( and not to internal errors of the method , as in table [ marvels_cand_param ] ) . the distribution of parameters for the thirty stars in the validation sample is also presented in figure [ samples ] ( filled squares ) . the low resolution spectra for the validation sample were obtained , as part of the survey , with the marvels instrument ( @xcite ; @xcite ) coupled to the sdss 2.5 m telescope at apo ( @xcite ) . the marvels instrument is a 60 object , fiber - fed , dispersed fixed delay interferometer ( dfdi ; @xcite ) that outputs two fringing spectra ( `` beams '' ) per object with a resolution r @xmath0 12,000 and a wavelength coverage between @xmath14 5000 - 5700 . since each star was visited typically @xmath5 20 times , the final combined spectra have s / n @xmath5 100 per pixel ( dispersion 0.154 / pixel ) . to convert the fringing spectra to conventional 1d extracted spectra , the former are first preprocessed , including corrections for optical distortion and slit illumination , creating a continuum - normalized fringing spectrum . then , for each wavelength , a sinusoid is fit to the fringing pattern which lies perpendicular to the wavelength axis . the dc offset of this sinusoid is the flux value of the normalized 1d extracted spectrum for that wavelength . these fluxes ( counts per pixel ) are converted into flux densities ( counts per unit wavelength ) , and then a barycentric rv correction is applied such that all 1d extracted spectra for a given star are registered to the rest wavelength . the determination of atmospheric parameters from these low - resolution marvels spectra and spectral indices is described in the following sections . the ability of the method to accurately recover the high - resolution results previously derived for the marvels stars is regarded as the validation test of the approach presented in this paper . we note that , in spite of its relatively small size , the adopted validation sample offers the most realistic check for the quality and reliability of our method since it uses real marvels data ( which includes instrumental effects , noise , etc ) . moreover , the set covers most regions of the parameter space defined by the calibration sample ( see figure [ samples ] ) and also the ranges of atmospheric parameters expected for the whole marvels sample ( according to the selection criteria employed during the survey target selection ) . as already mentioned , the marvels resolution prevents the characterization of the stars through measurements of equivalent widths of individual lines ( and consequently classical model atmosphere analysis ) because most of the spectral lines become blended with neighboring features . to overcome this problem , we have used instead an approach based on spectral indices ( e.g.,@xcite ) , which are defined here as groups of lines formed by similar chemical species . more specifically , we have selected two groups of interest with features dominated by : ( 1 ) neutral iron - peak species ( such as fe i , v i , cr i , mn i , co i and ni i ) and ( 2 ) ionized species ( such as fe ii , ti ii , and cr ii ) . these groups have , in principle , properties that should allow us to constrain , [ fe / h ] and . note that we consider a feature to be dominated by a certain group of elements when their respective lines account for more than 90% of its total absorption . the two groups were identified through a detailed inspection of the interval 5100 - 5590 on two versions of the same feros ganymede spectrum ( @xcite ) that was adopted as a solar template ( see figure [ indices_plot ] ) . the first version retained the original resolution ( r @xmath0 48,000 ) and was used in conjunction with @xcite to identify the individual lines that composed the indices . the second version was degraded to the marvels resolution and sampling ( see section [ calibsample ] ) and was used to determine the initial and final wavelengths of the selected indices . degraded feros spectra of the stars hd 32147 and hd 52298 were also utilized during the selection of the indices ( see figure [ indices_three_stars ] ) . hd 32147 is a cool metal - rich star ( = 4850 k and [ fe / h ] = 0.25 ; @xcite ) and corresponds to an extreme example of a spectrum with very strong indices . the analysis of its spectrum allowed us to refine the wavelength intervals previously defined for the indices ( based on the solar spectrum ) and to eliminate those that were not sufficiently isolated ( i.e. , did not have clear apparent continuum regions between them and the neighboring features ) . hd 52298 is a hot metal - poor star ( = 6253 k and [ fe / h ] = @xmath80.31 ; @xcite ) and illustrates the opposite case , in which the strengths of the indices become too weak , preventing accurate measurements of their equivalent widths . in summary , these two stars probe the extremes of the anticipated absorption strengths of the spectral indices and helped us to exclude possibly problematic features . following the procedure described above , a total of 96 potential indices were selected : 80 dominated by neutral iron - peak species and 16 by ionized species . the properties of these indices are listed in table [ indices_properties ] . we have referred to them as potential indices because their actual sensitivities to the atmospheric parameters are only analyzed in section [ regressions ] . the final list with the best available indices for the subsequent characterization of the marvels sample resulted from this analysis . the marvels sample contains @xmath03,300 stars with a high enough number of visits for robust planet detection and the time required to manually analyze all objects , each with @xmath540 spectra , would be prohibitively large . moreover , such an approach would be more error prone due to subjective choices that are inevitably made throughout the analysis of stellar spectra ( e.g. , during the continuum normalization ) . to avoid these issues , the determination of the atmospheric parameters was automatized through the development of three codes , which will be made publicly available throught the brazilian participation group scientific portal . these codes , as well as the tests performed to ascertain their quality , are described in more detail in the following sections . the normalization of a spectrum is a time - consuming task since one must test several parameters for choosing the curve that best describes the continuum of a particular star . as we aim to apply the indices technique for large stellar samples ( in particular , from the marvels survey ) , an automatic tool to perform this task is required . to this end , the first program was developed to fit a curve to and normalize the continuum on the spectra of solar - type stars . it was inspired by the task ` continuum ` from iraf , but has the advantage of automatically testing many different normalizations and determining which one provides the best fit . this feature turns the normalization process into a fast , non - interactive and systematic procedure , characteristics that are essential for the analysis of large samples of stars . the code uses as input the reduced , defringed , 1d doppler - corrected spectra ( in fits format ) that are given as part of the final products delivered by marvels pipeline . it is possible to combine more than one spectrum using the median counts as weights and cosmic rays can be removed through a @xmath15 clipping algorithm if at least five spectra are being co - added . a number of 1d legendre polynomials are then fit to the continuum points of the individual or combined spectra , in a wavelength range defined by the user . both the number of polynomials tested and of points considered in the fit depend on an additional set of input parameters provided by the user : polynomial order , high and low rejection limits ( as well as the steps with which they are changed ) , grow parameter and number of iterations to be performed . all parameters have the same definitions as in the task ` continuum ` from iraf . the program computes as many solutions as the choice of the input parameters described above , but testing all possible combinations for each star would require an excessive amount of time and most of the solutions would not be considered satisfactory . thus , we decided to restrict the set of input parameters using 27 stars ( seven with degraded feros spectra and 20 marvels targets with spectra from the survey ) with parameters that sample the range of indices strengths expected to be found in the stars from the marvels survey . we note that this test set included cool metal - rich stars to ensure that suitable apparent continuum regions could still be found , allowing accurate normalizations . the best visual normalizations were found for the following values of the input parameters : 5 or 6 for the order of the legendre polynomials ; 1.0 to 1.5 and 3.0 to 4.0 ( both with steps of 0.5 ) for the low and high rejection parameters ; 1 for the grow parameter ; 11 for the number of iterations . it was also observed that the fits with the highest correlation coefficient @xmath16 always produced the best results ( the same was not true for the cases with the lowest values of the standard deviation of the residuals @xmath15 ) and this is the reason why we chose this statistical criterion to select the final solutions . there are some limitations that could not be circumvented by the code after these restrictions were made . these issues are mainly related to the inability to deal with cosmic rays ( due to inefficient clipping or insufficient number of spectra to be combined ) or regions with defects ( e.g. , strong curvatures on the edges ) . the former can be minimized by combining a larger number of spectra for each star ( which was not possible for the stars with feros data ) . the second limitation is more complicated , but the usage of higher values for the polynomial order proved to be effective in most of the cases . the spectra of the 309 stars from the calibration sample ( section [ calibsample ] ) were normalized using the code and the restrictions and criteria described above . an example of the normalization procedure is shown in figure [ examples_norm ] . we visually inspected the final solutions for each of the 309 stars to search for normalization problems that could undermine the indices calibrations . the solutions provided by the code had to be replaced by visually better fits in 9% of the cases , which defines a reasonable limit to the degree of efficient automation of the normalization process . for the analysis of the complete marvels sample , the number of spectra will be larger by a factor of @xmath0100 and such a visual inspection will not feasible . we are currently working on a complementary code to automatically identify these cases that require human intervention based on large variations of the continuum fit parameters or the measured ews for some indices . even though an effort was made to ensure that the continuum placement was as accurate as possible , it is possible that small offsets are still present for a few spectra . the typical s / n of the individual marvels spectra for the validation sample is @xmath0100 , which corresponds to a rms error of the continuum of @xmath0(s / n)@xmath170.01 . using the method described in sections [ ews ] , [ regressions ] and [ method ] as well as the final results presented in section [ finalresults ] , we have checked that an offset of @xmath131% in the continuum placement causes the following average variations on the derived atmospheric parameters : @xmath12= @xmath85 @xmath13 42 k , @xmath12[fe / h ] = @xmath130.12 @xmath13 0.04 and @xmath12= @xmath180.07 @xmath13 0.07 . note that these values correspond to the unlikely case in which a global systematic offset would shift all the spectra of all the stars in the sample in the same direction and by the same amount . therefore , they should be regarded as the maximum possible errors that could be introduced by normalization issues . the actual errors will therefore be smaller than estimated here ( see also discussion in section [ finalresults ] ) . the normalized spectra obtained above for the calibration stars were used as input for the next code , which measures the equivalent widths ( ews ) of a list of indices provided by the user by direct integration of their profiles . this method was chosen because most of the indices are formed by multiple lines , each with a different type of profile ( gaussian or voigt ) . the integrations were performed in the wavelength intervals defined for each index ( see table [ indices_properties ] ) adopting the value of 1.0 for the normalized continuum flux . this code was used to measure the equivalent widths of the 96 indices ( see section [ indices ] ) in the degraded feros spectra of all 309 stars from the calibration sample . we have checked that all indices ews could be correctly measured even for the most metal - poor stars in the sample . a quick look at columns `` ewmin '' in table [ calibrations_ew ] and `` notes '' in table [ indices_properties ] reveals that 4.81 m is the lowest value for a equivalent width used in the calibrations that were adopted for the derivation of the atmospheric parameters ( more details about this selection are given in the next section ) . this value occurs for the star hd 76932 (= 5850 k and [ fe / h ] = -0.84 ) in the calibration for index 80 and is consistent with the lower limit of the measuring capabilities of our automated software . problematic ews were identified and discarded during the construction of the regressions ( see section [ regressions ] ) , since bad measurements appeared as clear outliers relative to the general trends observed . for the marvels targets , multiple spectra from different visits are available for a given star and the outliers can be excluded by a simple @xmath15 clipping procedure . the previous section discussed the measurements of the equivalent widths of the 96 spectral indices in the spectra of the calibration stars . it was also mentioned in section [ calibsample ] that all these objects have precise atmospheric parameters ( , [ fe / h ] and ) derived from detailed and homogeneous analysis of high - resolution spectra . with both the ews and atmospheric parameters in hand , we explored the relations between these quantities through a multivariate analysis and the final result was a set of calibrations that allowed the subsequent characterization of the marvels validation sample based solely on spectral indices ( without any other priors ) . figure [ index_sensitivity ] shows an example of the behavior of the indices ews as a function the atmospheric parameters for the calibration sample . clear relations are readily visible for and [ fe / h ] , while only scatter can be seen for . although this initial qualitative analysis revealed some promising indices , it can be misleading for the cases which lack evident correlations . the scatter might result from a truly weak sensitivity of the index relative to the parameter in question , but can also be caused a stronger sensitivity to one of the other two parameters ( recall that these plots show only a 1d projection of a possibly 3d relation ) . problems during the measurement of the ews for particular indices ( e.g. , possible contaminations having larger contributions than anticipated ) certainly also play a role . therefore , we decided not to remove any of the indices based on this initial visual inspection . the starting point of the quantitative analysis was the choice of an appropriate model to describe the relations observed in plots similar to figure [ index_sensitivity ] . in principle , all indices ews depend on the three atmospheric parameters in a complicated way which includes the interdependency between these parameters . tests with different polynomial orders revealed that a quadratic model would be able to correctly describe most of the observed behaviors . thus , we decided to search for the best calibrations for each index using second - order polynomials with the following structure : @xmath19 } + c_{2}t_{\rm eff } + c_{3}\log g + \nonumber \\ c_{4}{\rm [ fe / h]}t_{\rm eff } + c_{5}{\rm [ fe / h]}\log g + c_{6}t_{\rm eff}\log g + \nonumber \\ c_{7}({\rm [ fe / h]})^{2 } + c_{8}(t_{\rm eff})^{2 } + c_{9}(\log g)^{2}\end{aligned}\ ] ] these models ensure that any eventual interdependence between the atmospheric parameters are taken into account . some of the terms , however , could be statistically insignificant and , to evaluate this possibility , the choice of the final best calibrations followed an iterative procedure , which was performed with a python routine developed for this work . the starting point was equation ( [ ewcalib ] ) , for which the coefficients @xmath20 , ... , @xmath21 were determined through the ordinary least squares ( ols ) method ( i.e. , without any weights ) . we have designated these the complete models . when the best fit was found , the outliers were removed with a 2@xmath15 clipping . the previous steps were repeated three times and this limit was chosen to avoid an excessive exclusion of points . the mean number of stars that appeared as outliers was 37 , with minimum and maximum values of 19 and 49 , respectively . these numbers correspond to @xmath06 - 16% of the complete calibration sample , which is a reasonable fraction . we have checked that these outliers do not correspond to the stars that had their normalizations replaced by visually better fits ( see section [ norm ] ) . at the end of this iteration , the following statistical quantities were obtained for the current model : the correlation coefficient @xmath16 of the fit , the standard deviation of the residuals ( @xmath15 ) and the bayesian information criterion ( bic ) . the code then searched for models that had the same polynomial order and a similar ability to fit the data , but with fewer terms ; these are the reduced models . the progression of these tests was hierarchical , i.e. , the higher - order terms were the first ones to be removed . in each trial , a temporary reduced model and its associated statistical quantities ( @xmath16 , @xmath15 and bic ) were calculated . this reduced model was then compared to the complete one and the temporarily removed term was definitely excluded from the polynomial if three criteria were simultaneously met : both @xmath15 and bic are lower for the reduced model and the corresponding coefficient @xmath22 is statistically insignificant according to the f - statistics . if these conditions were satisfied , the reduced model replaced the complete one and new trials were done until all terms were tested . if one of the conditions was not satisfied , the tested term returned to the model and the trials continued until the relevance of all terms was evaluated . after the iterative procedure described above , a final optimized calibration was obtained for each index . table [ calibrations_ew ] presents the information for the calibrations obtained for the total set of 96 indices . if no number is provided for a given coefficient , the corresponding term was excluded according to the tests of statistical significance explained above . besides @xmath16 and @xmath15 , the number of stars used in the fit and the validity ranges for the equivalent widths are also shown . the validity ranges for , [ fe / h ] and are not given because they are very similar to the parameter space covered by the calibration sample ( see section [ calibsample ] ) . in order to guarantee the determination of precise atmospheric parameters , we have decided to keep in the analysis only those indices for which the respective calibrations had @xmath23 0.9 . this cut removed four indices ( marked with a number 2 in the notes of table [ indices_properties ] ) : 1 , 41 , 64 and 65 . figure [ calib_example ] shows index 38 as an example of the final calibrations adopted in this work . the next step was the analysis of the residuals ( i.e. , the differences @xmath12ew between the equivalent widths ew@xmath24 calculated with equation [ ewcalib ] and the equivalent widths ew@xmath25 measured in the stellar spectra ) for the remaining 92 calibrations . these were plotted as a function of the variables ( ew , , [ fe / h ] and ) and linear fits were applied to the data to evaluate the existence of any systematics . an example is presented in figure [ residuals ] for the same index as in figure [ index_sensitivity ] . the average correlation coefficient @xmath16 for the fits in which the ew is the dependent variable is 0.024 @xmath13 0.017 , with minimum and maximum values 0.006 and 0.096 , respectively . for the cases with , [ fe / h ] and as the independent variables , the values of @xmath16 are essentially null for all indices . these results demonstrate that the residuals are free of any significant systematic trends and there is no need to apply any a posteriori linear corrections to the ews calculated with equation ( [ ewcalib ] ) . the last of the three codes developed for this work delivers the final products of our analysis , which are the atmospheric parameters and their associated uncertainties . to accomplish this task , the program only requires the equivalent widths and calibrations discussed in sections [ ews ] and [ regressions ] , respectively . these can be changed in order to account for differences in resolution and wavelength range of the spectra . the choice of the best atmospheric parameters for each star is based on the minimization of a reduced chi - square ( @xmath26 ) and is described below . first , a set of 92 theoretical ews ( one for each index ) is calculated for each point of a three - dimensional grid of atmospheric parameters . the grid covers the following intervals : 4700 k @xmath1 @xmath1 6600 k , with 10 k steps ; @xmath80.90 @xmath1 [ fe / h ] @xmath1 0.50 , with 0.02 dex steps ; 3.50 @xmath1 @xmath1 4.70 , with 0.05 dex steps . these ranges are consistent with the parameter space covered by the calibration sample . smaller steps were tested , but no improvements in the results were observed . then , for each set of atmospheric parameters , a comparison between the observed and theoretical ews is performed through the calculation of a reduced chi - square as follows : @xmath27 in the above equation , @xmath28 and @xmath29 are the equivalent widths of the @xmath30-th index measured on the observed spectra and calculated with the calibrations , respectively . the number of indices used is given by @xmath31 . finally , @xmath32 is an error associated with the equivalent widths . for the calibration sample , it is simply the standard deviation found for each calibration , i.e. , @xmath33 . for the marvels validation sample , @xmath32 is given by the quadratic sum of this term and the standard deviation of the average ews measured on the stellar spectra , i.e. , @xmath34 ( more details in section [ marvelsresults ] ) . the final atmospheric parameters are those that produce the minimum value of @xmath26 . this value is also used for the determination of the associated uncertainties . first , we consider all sets of parameters that have @xmath35 , then calculate the differences between these parameters and those considered to be the best ones ( corresponding to @xmath36 ) . finally , the root mean squares of these differences are taken as the errors of the parameters . the previous sections were devoted to a description of the method that was used to derive the atmospheric parameters based only on the equivalent widths of spectral indices . we now discuss the tests performed to show that this approach is capable of recovering the precise atmospheric parameters determined from high - resolution stellar spectra and model atmosphere analysis . the first test was a sanity check with the calibration sample . the same equivalent widths employed to build the calibrations were used to derive the atmospheric parameters for the 309 calibration stars . the comparison between these results and those derived from high - resolution analysis is presented in figure [ param_calib ] . the average differences obtained for each parameter are given in the upper part of table [ diff_param ] . the offsets between the two sets of parameters are negligible and the dispersions around the average residuals are lower or of the order of the external errors usually found in high - resolution analyses ( for instance , see table 5 of @xcite ) . the situation is similar for the internal uncertainties of the method ( based on the @xmath26 ; see section [ method ] ) , which have typical values in the ranges 50 150 k for , 0.05 0.10 dex for [ fe / h ] and 0.10 0.25 dex for . the obvious outlier in the middle row panels of figure [ param_calib ] is a star that was excluded for 94 indices during the regressive analysis ( see section [ regressions ] ) . no significant trends can be seen in the residuals , except for systematically higher and lower values at @xmath05400 k and above @xmath06300 k , respectively . the latter behavior is probably caused by the reduced number of calibration stars ( only seven ) or by the fact that most of the indices become weaker in this temperature interval . this hypothesis is further supported by the behavior of the internal uncertainties ; slight increases can be observed towards higher effective temperatures or lower metallicities . in both regimes , we have fewer stars and smaller ews . the behavior at @xmath05400 k could be related to the discontinuity in the number of calibrations stars with cooler and hotter effective temperatures ( see figure [ samples ] ) . this systematic effect , however , has a small amplitude , with the majority of the residuals being within @xmath13100 k. as a complementary check , we have also analyzed the feros solar spectrum from @xcite with the same approach as above . the derived atmospheric parameters were : = 5720 @xmath13 95 k , [ fe / h ] = @xmath80.02 @xmath13 0.06 and = 4.30 @xmath13 0.16 . we can see that there is a good agreement with the standard solar parameters within the uncertainties . this result is expected given that the selection of the indices was based on a ganymede spectrum used as solar template . the above results thus provide a confirmation that the spectral indices method produces internally consistent atmospheric parameters . the second test had the goal of demonstrating that the method is capable of accurately recovering the high - resolution atmospheric parameters of a given stellar sample that was not used in the calibration of the method . to perform this exercise , we used the validation sample of 30 stars observed as part of the marvels survey ( see section [ testsample ] ) . the approach described in section [ pipeline ] was used to obtain the atmospheric parameters from the reduced , defringed , 1d , doppler - corrected marvels spectra . a few remarks are needed , however . all marvels stars have two sets of spectra because their light is divided into two `` beams '' that are collected by two neighboring fibers ( see section [ testsample ] ) . each set of spectra was analyzed separately and the exposures within each group were not combined for the normalization procedure . this procedure was done because we wanted to have multiple measurements for the ews of the indices in order to remove possible outliers and to have an estimate of their uncertainties . the average ews obtained after an iterative 2@xmath15 clipping ( until there were no outliers left ) were used as input to the atmospheric parameters determination method . the @xmath15 used in the calculation of @xmath26 was given by the quadratic sum of @xmath37 and @xmath38 , as previously mentioned in section [ method ] . finally , the two sets of parameters for each star were combined using a simple arithmetic average , with the uncertainties being obtained through an error propagation . the results in this initial attempt were not acceptable . the offsets relative to the high - resolution parameters and the dispersions were too large ( see the upper part of table [ diff_param ] ) . moreover , there were significant systematic trends as a function of . the only possible causes for these issues were the two types of input data used in our method : the measured ews or the calibrations . however , the good results obtained from the test with the calibration sample ruled out the latter . direct comparisons between the equivalent widths measured in the feros degraded and marvels solar spectra revealed significant differences . however , these differences were not caused by problems in the normalization and did not present any systematic behavior as a function of @xmath39 , , [ fe / h ] , and the ews themselves . therefore , we decided to derive individual corrections for each index to place the marvels ews onto the feros scale . in order to derive the corrections , we have considered 120 marvels solar spectra ( one for each fiber of the instrument ) observed simultaneously on 17 november 2009 . there is a different wavelength solution for each fiber , so the ranges covered vary from one spectrum to another . to avoid having indices with significantly fewer equivalent width measurements than others , we have decided to use only the wavelength range that is present in all 120 expsoures , which corresponds to the interval @xmath05137 - 5543 . as a consequence of this choice , 18 indices were removed from our analysis : 2 - 7 on the blue end ( note that 1 was already excluded because its calibration had @xmath40 0.9 ) and 85 - 96 on the red end ( all marked with a number 3 in the notes of table [ indices_properties ] ) . for the remaining 74 indices , the average ews from the 120 solar spectra and their respective standard deviations were calculated . no clipping was performed because we wanted to retain all the information regarding the variations of the ews with the fiber number in the instrument . this is important because the marvels stars could have been observed with any of the 120 fibers . we then calculated the differences @xmath41 . if these differences were higher than the corresponding values for @xmath42 , the indices were removed . this was the case for the following ten indices ( marked with a number 4 in the notes of table [ indices_properties ] ) : 33 , 34 , 39 , 40 , 54 , 55 , 61 , 62 , 66 , 67 . finally , it was decided that the remaining 64 indices would have their average marvels equivalent widths corrected by the difference @xmath43 . although this correction was based only on the sun , it was applied to all stars in the validation sample , regardless of their parameters . as it is shown below , the correction proved to be effective . another constraint implemented for the analysis of the marvels stars was the removal of corrected average ews that were outside the ranges defined by each of the calibrations ( see table [ calibrations_ew ] ) . this cut was done to avoid any extrapolation outside the parameter space defined by our calibration sample . using the method with the above optimization to the marvels data , we rederived the atmospheric parameters for the 30 stars in the marvels validation sample . recall that the final results for each object were obtained from an arithmetic average of the parameters produced by its two associated fibers . the two sets of results are in good agreement , with average differences of @xmath818 @xmath13 91 k for , 0.01 @xmath13 0.06 dex for [ fe / h ] and @xmath80.09 @xmath13 0.16 dex for . these values highlight that the data from adjacent fibers on the marvels instrument are consistent . the comparison between the atmospheric parameters derived from the indices and high - resolution analyses is presented in figure [ param_marvels ] . the average differences obtained for each parameter are given in the lower part of table [ diff_param ] . the offset in the metallicity is once again close to zero . for , we have a non - negligible negative value , but it is lower than the dispersion of the residuals and the typical uncertainties found for this parameter . a similar situation was found for . however , the offset in this case can be traced back to the reference values for the surface gravity . we recall that the values for the calibration sample present a small offset relative to other literature values ( see table 5 of @xcite ) . the dispersions around the average offsets are either lower than , or on the order of , the typical external uncertainties of high - resolution analyses . this was also the case for the internal uncertainties of the method , which have typical values in the ranges 80 160 k for , 0.05 0.10 dex for [ fe / h ] and 0.15 0.25 dex for . all these values are close to the ones derived for the calibration sample . the residuals do not exhibit any significant trends , except possibly for the same ones observed for in the results for the calibration sample . the residuals and internal uncertainties do not exhibit significant trends as a function of the s / n values of the spectra or the @xmath44 magnitudes of the stars ( taken from the guide star catalog ; @xcite ) , as can be seen on figure [ delta_sig_sn ] . the total s / n for a given star was obtained from the arithmetic average of the values derived from its two sets of spectra ( one from each fiber ) . the s / n for each set was calculated by multiplying the average s / n of the individual spectra @xmath45 by the square root of the number of spectra @xmath46 . this is a reasonable approximation considering that the exposure times are similar for all spectra of a given star . the lack of a dependency between the accuracy and precision ( differences and errors in left and right columns of figure [ delta_sig_sn ] , respectively ) of the results and the s / n of the spectra can be explained by a particular characteristic of our data . all stars have s / n @xmath47 200 per pixel ( dispersion 0.154 / pixel ) and this value is certainly higher than the limit below which we would start to observe a decline in the quality of the results with decreasing s / n . thus , we conclude that all stars have high - quality data that allow our approach to work at its best precision and accuracy . the method was further tested using the 120 marvels solar spectra described previously . since we wanted to obtain the best possible results , an iterative 2@xmath15 clipping ( until there were no outliers left ) was applied to calculate the average ews of the sun . following the exact same method employed for the marvels validation sample , we derived the results : = 5760 @xmath13 81 k , [ fe / h ] = @xmath80.04 @xmath13 0.05 and = 4.40 @xmath13 0.15 . the agreement with the standard solar parameters is excellent . for completeness , we also utilized the restricted set of 64 indices ( without any corrections ) to rederive the parameters for the calibration stars . as can be seen on the lower part of table [ diff_param ] , there are no significant differences relative to the previous case in which all 92 indices were considered . the above results for the marvels validation sample show that the spectral indices approach is able to accurately recover the precise atmospheric parameters derived from the analysis of high - resolution spectra . moreover , the differences between the two sets of results and the internal uncertainties of the method presented here are both consistent with the typical external errors obtained in high - resolution analyzes . it should be highlighted that this is achieved by using only equivalent widths directly measured on the observed low - resolution marvels spectra , with no other priors . the third and last test was conducted to complement the previous one . although the validation sample contains stars with real spectra from the marvels survey , we recognize that 30 stars is a relatively small number to critically evaluate the performance of our method . therefore , we decided to analyze the spectra contained in the elodie stellar library ( @xcite , @xcite ; see website for the most updated 3.1 version ) . briefly , the library includes 1962 high - resolution ( r = 42,000 ) spectra observed with elodie spectrograph coupled to the 1.93 m telescope at observatoire de haute - provence . the spectra cover the wavelength range 3900 - 6800 , have s / n values varying between 30 and 680 and belong to 1388 stars that non - uniformly sample the following intervals of atmospheric parameters : 3442 k @xmath1 @xmath1 47250 k , @xmath82.94 @xmath1 [ fe / h ] @xmath1 @xmath91.40 and 0.00 @xmath1 @xmath1 4.90 . the above values are quoted from the source catalog for reference , but our analysis is restricted to only those stars who have literature average atmospheric parameters within the limits defined by our calibration sample ( 4800 k @xmath7 @xmath7 6500 k , @xmath80.90 @xmath7 [ fe / h ] @xmath7 @xmath90.50 and 3.60 @xmath7 @xmath7 4.70 ) . we have also restricted our selection to stars with classes 3 or 4 attributed to their average parameters , which means that the typical standard deviations around their mean and [ fe / h ] values are 62 k and 0.08 dex and 74 k and 0.10 dex , respectively . no cuts in s / n were applied , but stars with indications of variability or close neighbors were removed . the final elodie test sample consists of 219 spectra from 138 unique stars . as for the feros sample , these spectra were degraded to the marvels resolution following the same procedure described in section [ calibsample ] . the elodie test sample was analyzed with the approach presented in section [ pipeline ] . all spectra were normalized individually and the equivalent widths of the indices were measured in each of them , as if they belonged to different stars . this was done to check the internal consistency of the atmospheric parameter determinations for a given star based on different spectra , which found to be good . the ews of the 92 indices for which the calibrations have r@xmath48 0.9 were considered . again , a data - specific tuning was necessary and a correction to the equivalent width scale based on the sun was performed . average solar elodie equivalent widths were calculated using 11 of the 13 spectra of the sun with s / n @xmath49 100 . the differences between these average ews and the values measured on the feros solar degraded spectrum were adopted as individual corrections for each index . using these corrected equivalent widths and the calibrations presented in section [ regressions ] , atmospheric parameters were derived for the elodie sample . the results can be seen in figure [ param_elodie ] and in table [ diff_param ] . we can see that the agreement for and [ fe / h ] is good and the average differences and dispersions ( 15 @xmath13 125 k and 0.06 @xmath13 0.10 , respectively ) can be entirely explained by a combination of the internal and external uncertainties from our method with the typical standard deviations around the mean elodie parameters . mild trends are observed for the residuals of these parameters , with the differences between the two sets of results increasing for hot and metal - poor stars , which is expected due to the lower number of calibration stars in these regions of the parameter space . spectra with low s / n values also contribute for the observed trend , since they are clearly associated with the largest residuals ( see also figure [ delta_sig_sn_elodie ] ) . the agreement for , on the other hand , is only reasonable , with a relatively large dispersion and clear trends in the residuals . this result was not totally unexpected since the previous tests with the calibration and validation samples have shown that the surface gravity is the most difficult parameter to accurately constrain . although the average difference and dispersion ( 0.07 @xmath13 0.29 ) can again be explained by a combination of the uncertainties in the two sets of parameters , we can see that some stars have differences higher than 0.5 dex . however , it is clear from the lower panels that these stars are hot and/or metal - poor . thus , the poor determinations of and [ fe / h ] could also be affecting the estimate of . many of these stars also have spectra with low s / n values ( see figure [ delta_sig_sn_elodie ] ) . the average solar parameters derived from the 13 spectra available are : = 5750 @xmath13 66 k , [ fe / h ] = 0.00 @xmath13 0.04 and = 4.32 @xmath13 0.11 . the agreement with the canonical solar parameters = 5777 k , [ fe / h ] = 0.00 and = 4.44 ) is very good , with the surface gravity providing the most discrepant result once again . the elodie test sample also offers the opportunity to check the performance of our method for lower s / n spectra , which was not possible for the marvels validation sample . figure [ delta_sig_sn_elodie ] shows the differences between the spectral indices and elodie atmospheric parameters as well as the uncertainties of the former as a function of the s / n per pixel of the spectra . for s / n @xmath49 200 - 250 , the typical values of the differences and uncertainties agree well with the ones for the marvels sample . at s / n @xmath0 200 - 250 , there is a significant increase in both quantities . for s / n @xmath50 200 - 250 , the scatter of the distributions stay pretty much constant . these results show that our method is capable of providing reliable results for spectra with lower s / n , although with lower accuracies than in the high - s / n regime . the tests discussed above provide an additional validation of the spectral indices method , based on a larger sample of observed stellar spectra with typically lower s / n values . as in the previous sections , the precise atmospheric parameters from the literature were recovered with a good precision . the small trends observed in the residuals are caused by the non - uniform sampling of the parameter space by our calibration sample and the lower sensitivities of the indices to the surface gravity . in this paper , we described the usage of spectral indices as a fast and homogeneous approach to determine accurate atmospheric parameters ( , [ fe / h ] , ) for samples of fgk dwarfs and subgiants with intermediate ( r @xmath0 12,000 ) resolution spectra , applied to observations from the marvels survey . equivalent widths of the indices were measured on normalized spectra and then compared to values calculated with a set of calibrations , which were constructed using 309 stars with precise atmospheric parameters resulting from the analysis of feros spectra . the best solutions were obtained through a @xmath26 minimization of the differences between these two sets of ews . the entire analysis was automatized with three codes developed for this work . the spectral indices method was validated with a sample of 30 stars which were observed as part of the marvels survey and have precise atmospheric parameters derived from classical model atmosphere analyses . we were able to recover these parameters within 80 k for , 0.05 dex for [ fe / h ] and 0.15 dex for . the average internal errors of the method are 101 k , 0.06 dex and 0.15 dex , respectively . these values are consistent with the typical external uncertainties found between different high - resolution spectroscopic analyses ( see , e.g. , table 5 of @xcite ) . further confirmation of the good performance of the method was provided by the analysis of a subsample of 138 stars from the elodie stellar library with average literature atmospheric parameters with good quality . these were recovered by our method with precisions of 125 k for , 0.10 dex for [ fe / h ] and 0.29 dex for . the higher uncertainties are caused by a combination of the internal uncertainties of the spectral indices approach , the typically lower s / n values for the elodie spectra and the typical standard deviations around the mean elodie parameters . undersampled regions in the parameter space for the calibration sample might also play a role . however , these errors are still similar or lower than typical uncertainties provided by the pipelines of other intermediate - resolution surveys ( e.g. , rave ; @xcite ) . this agreement demonstrates that the spectral indices approach is a powerful tool to derive accurate and precise atmospheric parameters for solar - type stars . the method is quite general and its application to a particular data set requires only a small tuning of the input parameters ( list of indices and the measured equivalent widths ) . the approach has an excellent performance , yielding typical stellar parameter uncertainties comparable to or better than those provided by pipelines that utilize spectral synthesis , such as the ones from the segue , rave and lamost surveys . we note , however , that they analyze spectra with lower resolutions , s / n values , wavelength coverages or combinations of these properties . our method could be adopted by similar surveys , with modifications to the list of indices and calibrations as appropriate for differences in the resolution and wavelength coverage . the next step of our work is the application of the spectral indices method to the entire marvels sample ( @xmath03,300 stars ) , which contains many faint stars ( @xmath44 @xmath0 10 - 12 ) that were never previously analyzed . the atmospheric parameters will then be used to derive additional stellar properties , such as mass , radius , distances and ages . the kinematic analysis , on the other hand , will follow from the precise rvs determined by marvels . with accurate chemical and kinematical information for a large and statistically homogeneous sample , marvels can provide valuable contributions to many studies , such as the comparison of the statistical properties of stars with and without companions , the search for correlations between the properties of the companions and their stellar hosts , and galactic chemical and dynamical evolution in the solar neighborhood . these will be the subjects of subsequent papers . funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s . department of energy office of science . the sdss - iii web site is http://www.sdss3.org/. sdss - iii is managed by the astrophysical research consortium for the participating institutions of the sdss - iii collaboration including the university of arizona , the brazilian participation group , brookhaven national laboratory , carnegie mellon university , university of florida , the french participation group , the german participation group , harvard university , the instituto de astrofisica de canarias , the michigan state / notre dame / jina participation group , johns hopkins university , lawrence berkeley national laboratory , max planck institute for astrophysics , max planck institute for extraterrestrial physics , new mexico state university , new york university , ohio state university , pennsylvania state university , university of portsmouth , princeton university , the spanish participation group , university of tokyo , university of utah , vanderbilt university , university of virginia , university of washington , and yale university . funding for the brazilian participation group has been provided by the ministrio de cincia e tecnologia ( mct ) , fundao carlos chagas filho de amparo pesquisa do estado do rio de janeiro ( faperj ) , conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) , and financiadora de estudos e projetos ( finep ) . we thank eduardo del peloso for developing the codes used to normalize the spectra and measure the equivalent widths of the indices . we acknowledge katia cunha , verne smith and daniel eisenstein for helpful suggestions . we thank the anonymous referee for the thorough reading of the manuscript and the detailed comments which helped improving it . l.g . acknowledges financial support provided by the papdrj capes / faperj fellowship . l.g . thanks k. cunha and v. smith for discussions and for helping with the acquisition of the feros spectra . l.d.f . and d.l.o . acknowledge financial support from capes and eso student fellowships . g.f.p.m . acknowledges the financial support by cnpq ( 476909/2006 - 6 and 474972/2009 - 7 ) and faperj ( apq1/26/170.687/2004 ) grants . acknowledges support from cnpq ( 301462/2009 - 7 ) . zucker , d. b. , de silva , g. , freeman , k. , bland - hawthorn , j. , & hermes team 2012 , in asp conf . 458 , galactic archaeology : near - field cosmology and the formation of the milky way , ed . w. aoki , m. ishigaki , t. suda , t. tsujimoto , & n. arimoto ( san francisco , ca : asp ) , 421 lcrcc mc1 & 6297 @xmath13 28 & @xmath80.13 @xmath13 0.04 & 4.22 @xmath13 0.09 & @xcite + mc2 & 5598 @xmath13 63 & 0.40 @xmath13 0.09 & 4.44 @xmath13 0.17 & @xcite + mc5 & 6214 @xmath13 38 & @xmath80.45 @xmath13 0.06 & 4.38 @xmath13 0.15 & this work + mc6 & 6427 @xmath13 33 & @xmath80.04 @xmath13 0.05 & 4.52 @xmath13 0.14 & @xcite + mc7 & 5879 @xmath13 29 & @xmath80.01 @xmath13 0.05 & 4.48 @xmath13 0.15 & @xcite + mc11 & 6004 @xmath13 29 & 0.04 @xmath13 0.05 & 4.55 @xmath13 0.15 & @xcite + mc12 & 5903 @xmath13 42 & @xmath80.23 @xmath13 0.07 & 4.07 @xmath13 0.16 & @xcite + mckgs1 - 50 & 5540 @xmath13 39 & 0.20 @xmath13 0.07 & 4.48 @xmath13 0.23 & this work + mckgs1 - 52 & 6403 @xmath13 44 & 0.17 @xmath13 0.07 & 4.42 @xmath13 0.24 & this work + mckgs1 - 61 & 5757 @xmath13 43 & @xmath80.22 @xmath13 0.07 & 4.29 @xmath13 0.22 & this work + mckgs1 - 70 & 5135 @xmath13 39 & @xmath80.01 @xmath13 0.08 & 3.42 @xmath13 0.18 & this work + mckgs1 - 94 & 4903 @xmath13 47 & @xmath80.49 @xmath13 0.08 & 4.48 @xmath13 0.28 & this work + mckgs1 - 112 & 5782 @xmath13 42 & 0.01 @xmath13 0.07 & 4.21 @xmath13 0.23 & this work + mckgs1 - 135 & 5525 @xmath13 71 & @xmath80.17 @xmath13 0.08 & 4.39 @xmath13 0.23 & this work + mckgs1 - 153 & 5614 @xmath13 29 & @xmath80.09 @xmath13 0.06 & 4.61 @xmath13 0.09 & this work + mcuf1 - 11 & 5315 @xmath13 44 & 0.33 @xmath13 0.06 & 4.23 @xmath13 0.19 & this work + llrcc wasp 1 & 6161 @xmath13 52 & 0.19 @xmath13 0.06 & 4.23 @xmath13 0.05 & 1,2,3 + hd 4203 & 5644 @xmath13 61 & 0.41 @xmath13 0.03 & 4.24 @xmath13 0.14 & 4,5,6,7,8,9,10,11 + hd 9407 & 5661 @xmath13 5 & 0.03 @xmath13 0.02 & 4.45 @xmath13 0.03 & 12,8,13 + hd 17156 & 6057 @xmath13 46 & 0.19 @xmath13 0.05 & 4.20 @xmath13 0.11 & 1,14,7,15,16,17 + hip 14810 & 5515 @xmath13 25 & 0.27 @xmath13 0.02 & 4.27 @xmath13 0.06 & 18,14,5,7,19 + hd 43691 & 6200 @xmath13 39 & 0.29 @xmath13 0.02 & 4.23 @xmath13 0.12 & 14,18,7,20 + hd 49674 & 5632 @xmath13 31 & 0.33 @xmath13 0.01 & 4.48 @xmath13 0.12 & 18,7,8,9 + xo-2 & 5356 @xmath13 19 & 0.41 @xmath13 0.05 & 4.36 @xmath13 0.19 & 1,16,21 + hd 68988 & 5968 @xmath13 48 & 0.35 @xmath13 0.02 & 4.44 @xmath13 0.08 & 18,7,22,8,9,11,23 + hd 80606 & 5615 @xmath13 70 & 0.37 @xmath13 0.08 & 4.44 @xmath13 0.07 & 1,7,22,8,9,23,10,24 + hd 118203 & 5767 @xmath13 70 & 0.15 @xmath13 0.06 & 3.92 @xmath13 0.04 & 25,18,7 + hat - p-3 & 5205 @xmath13 28 & 0.34 @xmath13 0.10 & 4.60 @xmath13 0.02 & 1,26 + tres-2 & 5840 @xmath13 41 & @xmath80.03 @xmath13 0.11 & 4.39 @xmath13 0.09 & 1,16,27 + hat - p-1 & 6026 @xmath13 71 & 0.17 @xmath13 0.06 & 4.46 @xmath13 0.01 & 16,28 + ccccc 1 & 5100.40 & 5102.00 & fe ii & 2 + 2 & 5102.00 & 5104.90 & ni i , fe i & 3 + 3 & 5105.10 & 5106.05 & cu i & 3 + 4 & 5106.90 & 5108.25 & fe i & 3 + 5 & 5109.95 & 5111.05 & fe i & 3 + 6 & 5130.95 & 5132.20 & fe i , ni i & 3 + 7 & 5135.55 & 5137.90 & fe i , ni i & 3 + 8 & 5138.05 & 5140.45 & fe i , cr i & 1 + 9 & 5140.65 & 5144.15 & fe i , ni i & 1 + 10 & 5194.35 & 5196.95 & fe i , cr i , mn i & 1 + 11 & 5196.95 & 5198.00 & fe ii , fe i , ni i & 1 + 12 & 5198.25 & 5199.35 & fe i & 1 + 13 & 5201.55 & 5203.30 & fe i & 1 + 14 & 5203.30 & 5209.45 & cr i , fe i & 1 + 15 & 5213.55 & 5216.90 & fe i & 1 + 16 & 5229.10 & 5230.95 & fe i & 1 + 17 & 5231.85 & 5233.70 & fe i & 1 + 18 & 5234.05 & 5236.75 & fe ii , fe i & 1 + 19 & 5236.75 & 5238.00 & cr ii & 1 + 20 & 5241.25 & 5244.85 & fe i & 1 + 21 & 5249.75 & 5254.25 & fe i & 1 + 22 & 5254.25 & 5256.35 & cr i , fe i , mn i & 1 + 23 & 5272.60 & 5274.05 & fe i & 1 + 24 & 5274.05 & 5276.80 & fe ii , fe i , cr i , cr ii & 1 + 25 & 5279.35 & 5280.95 & cr i , fe i , cr ii & 1 + 26 & 5280.95 & 5282.85 & fe i & 1 + 27 & 5282.85 & 5286.65 & fe i , ti i , fe ii & 1 + 28 & 5288.00 & 5289.10 & fe i & 1 + 29 & 5296.15 & 5299.45 & cr i , fe i & 1 + 30 & 5301.50 & 5303.25 & fe i & 1 + 31 & 5306.75 & 5308.00 & fe i & 1 + 32 & 5308.00 & 5309.45 & cr ii & 1 + 33 & 5312.10 & 5314.25 & cr ii , cr i & 4 + 34 & 5314.25 & 5315.65 & fe i & 4 + 35 & 5315.90 & 5317.90 & fe ii & 1 + 36 & 5320.50 & 5322.65 & fe i & 1 + 37 & 5322.65 & 5325.00 & fe i & 1 + 38 & 5327.05 & 5330.85 & fe i , cr i & 1 + 39 & 5332.05 & 5333.55 & fe i , v ii & 4 + 40 & 5334.55 & 5335.60 & cr ii , co i & 4 + 41 & 5336.30 & 5337.35 & ti ii & 2 + 42 & 5338.90 & 5341.95 & fe i , cr i , mn i & 1 + 43 & 5341.95 & 5344.05 & co i , fe i & 1 + 44 & 5345.10 & 5346.40 & cr i & 1 + 45 & 5347.35 & 5348.90 & cr i & 1 + 46 & 5352.65 & 5354.20 & fe i , co i , ni i & 1 + 47 & 5360.70 & 5362.15 & fe i & 1 + 48 & 5362.20 & 5363.65 & fe ii , fe i , co i & 1 + 49 & 5363.65 & 5366.05 & fe i & 1 + 50 & 5366.90 & 5368.15 & fe i & 1 + 51 & 5368.95 & 5372.85 & fe i , ni i , co i & 1 + 52 & 5373.10 & 5374.70 & fe i & 1 + 53 & 5377.10 & 5378.55 & mn i & 1 + 54 & 5382.65 & 5384.35 & fe i & 4 + 55 & 5385.85 & 5388.00 & fe i , cr i & 4 + 56 & 5388.85 & 5392.30 & fe i & 1 + 57 & 5392.50 & 5394.05 & fe i & 1 + 58 & 5394.05 & 5395.75 & mn i , fe i & 1 + 59 & 5395.75 & 5398.95 & fe i & 1 + 60 & 5399.95 & 5402.05 & fe i , cr i & 1 + 61 & 5403.20 & 5407.00 & fe i & 4 + 62 & 5408.35 & 5411.80 & cr i , fe i , ni i & 4 + 63 & 5414.45 & 5416.00 & fe i & 1 + 64 & 5416.50 & 5417.60 & fe i & 2 + 65 & 5417.65 & 5419.60 & ti ii & 2 + 66 & 5422.90 & 5425.65 & fe i , ni i , fe ii & 4 + 67 & 5427.30 & 5431.00 & fe i & 4 + 68 & 5431.90 & 5435.35 & fe i , mn i & 1 + 69 & 5435.35 & 5437.60 & ni i , fe i & 1 + 70 & 5454.30 & 5456.95 & fe i & 1 + 71 & 5462.00 & 5465.20 & fe i , ni i & 1 + 72 & 5465.20 & 5467.55 & fe i & 1 + 73 & 5469.70 & 5471.75 & mn i , fe i & 1 + 74 & 5475.65 & 5477.45 & ni i , fe i & 1 + 75 & 5482.65 & 5484.25 & fe i , co i & 1 + 76 & 5486.35 & 5488.60 & fe i & 1 + 77 & 5496.80 & 5498.50 & fe i & 1 + 78 & 5500.20 & 5502.45 & fe i & 1 + 79 & 5505.10 & 5507.55 & fe i , mn i & 1 + 80 & 5516.00 & 5517.75 & mn i , fe i & 1 + 81 & 5521.85 & 5522.95 & fe i & 1 + 82 & 5524.70 & 5526.15 & fe i & 1 + 83 & 5531.45 & 5533.80 & fe i & 1 + 84 & 5534.05 & 5536.15 & fe i , ba i , fe ii & 1 + 85 & 5542.30 & 5544.65 & fe i & 3 + 86 & 5545.50 & 5547.85 & fe i , v i & 3 + 87 & 5552.90 & 5555.70 & fe i , ni i & 3 + 88 & 5559.40 & 5560.85 & fe i & 3 + 89 & 5562.05 & 5564.50 & fe i & 3 + 90 & 5564.85 & 5566.45 & fe i & 3 + 91 & 5566.80 & 5568.20 & fe i & 3 + 92 & 5568.80 & 5570.30 & fe i & 3 + 93 & 5571.95 & 5574.00 & fe i & 3 + 94 & 5575.40 & 5576.80 & fe i & 3 + 95 & 5577.90 & 5579.25 & ni i & 3 + 96 & 5585.85 & 5587.35 & fe i & 3 + crrrrrrrrrrrrrrr 1 & 1.75e+03 & 3.28e+02 & -1.04e+00 & 6.82e+02 & -1.17e-02 & -4.34e+01 & -5.57e-02 & 5.88e+01 & 1.06e-04 & -4.39e+01 & 0.895 & 11.55 & 261 & 1.65 & 230.01 + 2 & 7.65e+02 & 8.84e+02 & -5.20e-01 & 6.13e+02 & -1.97e-02 & -1.36e+02 & -2.72e-02 & 2.39e+01 & 4.35e-05 & -5.49e+01 & 0.926 & 18.39 & 264 & 25.01 & 402.49 + 3 & 1.04e+02 & 1.37e+02 & -1.23e-01 & 2.43e+02 & 1.94e-02 & -4.35e+01 & & 7.23e+00 & 6.21e-06 & -3.02e+01 & 0.935 & 6.06 & 271 & 26.44 & 166.78 + 4 & 2.02e+03 & 4.08e+02 & -7.82e-01 & 3.48e+02 & -4.32e-02 & -1.31e+01 & -5.91e-02 & & 8.09e-05 & & 0.974 & 7.48 & 266 & 143.32 & 357.77 + 5 & 1.68e+03 & 4.54e+02 & -5.41e-01 & 1.37e+02 & -4.75e-02 & -1.74e+01 & -3.68e-02 & 7.28e+00 & 5.32e-05 & 9.16e+00 & 0.980 & 5.56 & 266 & 94.09 & 320.62 + 6 & 1.24e+03 & 1.42e+02 & -2.62e-01 & -8.39e+01 & 3.16e-02 & -5.26e+01 & -2.75e-02 & -7.08e+00 & 2.87e-05 & 2.75e+01 & 0.955 & 6.90 & 267 & 55.36 & 250.87 + 7 & 3.60e+03 & 5.65e+02 & -1.15e+00 & 1.65e+02 & -1.52e-02 & -6.65e+01 & -8.45e-02 & 2.62e+01 & 1.20e-04 & 3.87e+01 & 0.964 & 13.58 & 270 & 135.31 & 591.45 + 8 & 7.51e+03 & 3.36e+02 & -2.94e+00 & 9.48e+02 & -6.03e-02 & 4.85e+01 & -1.47e-01 & -5.02e+01 & 2.85e-04 & & 0.986 & 15.23 & 266 & 188.07 & 956.46 + 9 & 8.65e+03 & 7.99e+02 & -2.87e+00 & 3.53e+02 & -3.08e-02 & -8.81e+01 & -1.49e-01 & -5.27e+01 & 2.83e-04 & 6.37e+01 & 0.975 & 20.20 & 266 & 233.52 & 1006.07 + 10 & 5.71e+03 & 2.62e+02 & -2.22e+00 & 7.34e+02 & -2.53e-02 & 2.64e+01 & -2.11e-01 & & 2.52e-04 & 6.17e+01 & 0.985 & 13.55 & 268 & 183.55 & 851.70 + 11 & 9.15e+02 & 7.32e+01 & -2.32e-01 & -4.08e+01 & 1.82e-02 & -2.51e+01 & -3.08e-02 & 4.12e+00 & 3.20e-05 & 2.26e+01 & 0.946 & 4.75 & 266 & 58.48 & 176.57 + 12 & 2.34e+03 & 3.41e+01 & -9.10e-01 & 2.41e+02 & & & -9.01e-02 & -1.30e+01 & 1.08e-04 & 3.38e+01 & 0.966 & 5.93 & 272 & 50.29 & 242.45 + 13 & 5.47e+03 & 7.34e+01 & -2.10e+00 & 4.87e+02 & & & -1.15e-01 & & 2.15e-04 & 2.24e+01 & 0.976 & 9.78 & 264 & 92.17 & 524.25 + 14 & 3.25e+04 & 5.86e+02 & -1.37e+01 & 4.71e+03 & -2.32e-01 & 2.90e+02 & -7.72e-01 & & 1.39e-03 & & 0.988 & 46.94 & 263 & 474.31 & 3092.88 + 15 & 4.08e+03 & 8.07e+02 & -1.76e+00 & 8.91e+02 & -1.21e-01 & 4.68e+01 & -1.85e-01 & 7.75e+01 & 2.02e-04 & 2.38e+01 & 0.985 & 13.75 & 260 & 169.39 & 773.83 + 16 & 1.71e+03 & 3.30e+02 & -7.10e-01 & 3.64e+02 & -6.17e-02 & 3.53e+01 & -8.26e-02 & 1.60e+01 & 8.21e-05 & 1.62e+01 & 0.987 & 6.05 & 266 & 69.96 & 360.47 + 17 & 1.90e+03 & 1.86e+02 & -1.37e+00 & 1.37e+03 & -7.87e-02 & 1.13e+02 & -1.50e-01 & -1.23e+01 & 1.51e-04 & -4.58e+01 & 0.991 & 9.96 & 262 & 149.67 & 708.59 + 18 & 2.20e+03 & 3.63e+02 & -2.68e-01 & -4.13e+02 & -1.13e-02 & -1.87e+01 & -2.94e-02 & 4.84e+01 & 2.78e-05 & 6.48e+01 & 0.982 & 8.22 & 271 & 126.15 & 473.52 + 19 & 9.82e+02 & 1.07e+02 & -1.69e-01 & -2.06e+02 & -3.79e-03 & -1.00e+01 & -7.31e-03 & 1.57e+01 & 1.89e-05 & 2.66e+01 & 0.934 & 3.45 & 269 & 32.07 & 99.20 + 20 & 1.16e+03 & 2.46e+02 & -3.95e-01 & 2.04e+02 & -8.85e-03 & & -3.79e-02 & 5.90e+01 & 4.20e-05 & & 0.968 & 9.63 & 265 & 88.92 & 400.24 + 21 & 7.79e+03 & 5.77e+02 & -2.11e+00 & -2.62e+02 & -7.50e-02 & 3.63e+01 & -3.08e-02 & 1.00e+02 & 1.72e-04 & 5.10e+01 & 0.987 & 13.15 & 269 & 183.32 & 846.77 + 22 & 4.74e+03 & 7.74e+02 & -1.24e+00 & -2.14e+02 & -9.45e-02 & & 1.57e-02 & 7.61e+01 & 8.84e-05 & 1.24e+01 & 0.991 & 7.21 & 268 & 94.78 & 541.35 + 23 & 1.93e+03 & 1.49e+02 & -6.57e-01 & 2.18e+02 & -3.25e-02 & 3.88e+01 & -6.54e-02 & 1.00e+01 & 7.16e-05 & 2.11e+01 & 0.988 & 5.30 & 267 & 131.04 & 403.42 + 24 & 3.01e+03 & 5.03e+02 & -8.45e-01 & 9.67e+01 & -4.19e-02 & 1.58e+01 & -4.80e-02 & 6.81e+01 & 8.12e-05 & 1.84e+01 & 0.986 & 10.74 & 271 & 201.06 & 667.33 + 25 & 1.35e+03 & 3.09e+02 & -3.59e-01 & 5.66e+00 & -8.67e-03 & -2.60e+01 & -6.37e-03 & 5.53e+01 & 2.95e-05 & & 0.988 & 4.40 & 267 & 47.65 & 265.59 + 26 & 2.97e+03 & 3.25e+02 & -1.21e+00 & 5.37e+02 & -8.44e-02 & 7.94e+01 & -6.54e-02 & 7.01e+01 & 1.15e-04 & -1.43e+01 & 0.990 & 7.13 & 262 & 110.44 & 517.87 + 27 & 3.42e+03 & 3.37e+02 & -1.34e+00 & 6.61e+02 & -7.75e-02 & 9.46e+01 & -5.72e-02 & 7.98e+01 & 1.19e-04 & -3.55e+01 & 0.989 & 10.87 & 264 & 210.44 & 742.70 + 28 & 1.08e+03 & 1.48e+02 & -2.64e-01 & -6.03e+01 & -1.54e-02 & & 9.59e-03 & 1.55e+01 & 1.62e-05 & & 0.987 & 2.08 & 263 & 33.97 & 135.33 + 29 & 6.57e+03 & 8.00e+02 & -2.22e+00 & 5.23e+02 & -1.02e-01 & 3.75e+01 & -2.14e-01 & 6.90e+01 & 2.45e-04 & 8.84e+01 & 0.991 & 13.90 & 270 & 232.40 & 1148.08 + 30 & 1.69e+03 & 2.85e+02 & -6.89e-01 & 3.40e+02 & -6.82e-02 & 5.45e+01 & -5.33e-02 & 3.26e+01 & 7.00e-05 & & 0.987 & 5.38 & 263 & 108.94 & 366.37 + 31 & 1.07e+03 & 3.13e+01 & -3.07e-01 & 6.39e+00 & & 5.64e+00 & -1.86e-02 & 1.09e+01 & 3.11e-05 & 1.10e+01 & 0.962 & 3.35 & 269 & 61.56 & 160.74 + 32 & 6.57e+02 & 2.07e+02 & -1.90e-01 & -1.62e+01 & -2.25e-02 & -7.30e+00 & & 2.89e+01 & 1.64e-05 & & 0.920 & 3.51 & 274 & 11.71 & 88.54 + 33 & 1.18e+03 & 2.70e+02 & -3.39e-01 & -2.06e+01 & -3.22e-02 & & & 4.33e+01 & 2.79e-05 & & 0.971 & 3.67 & 263 & 21.50 & 142.86 + 34 & 6.18e+02 & 2.40e+02 & -1.66e-01 & 4.39e+00 & -3.51e-02 & 6.30e+00 & 1.18e-02 & 3.14e+01 & 7.04e-06 & -8.84e+00 & 0.979 & 2.70 & 262 & 11.12 & 121.67 + 35 & 2.62e+02 & 2.06e+02 & 1.26e-01 & -2.02e+02 & -1.85e-02 & & 6.48e-03 & 2.52e+01 & -1.08e-05 & 1.49e+01 & 0.979 & 4.39 & 275 & 102.86 & 269.06 + 36 & 8.56e+02 & 9.15e+01 & -3.10e-01 & 1.85e+02 & -7.42e-03 & 1.45e+01 & & 1.52e+01 & 2.02e-05 & -2.22e+01 & 0.976 & 5.83 & 276 & 41.44 & 252.53 + 37 & 1.88e+03 & 2.31e+02 & -1.26e+00 & 1.23e+03 & -8.96e-02 & 1.21e+02 & -1.24e-01 & 2.13e+01 & 1.33e-04 & -4.83e+01 & 0.988 & 11.81 & 273 & 147.63 & 696.29 + 38 & 1.06e+04 & 5.17e+02 & -4.08e+00 & 1.52e+03 & -1.31e-01 & 1.59e+02 & -2.37e-01 & & 3.92e-04 & & 0.993 & 18.36 & 266 & 369.10 & 1453.86 + 39 & 1.77e+03 & 2.44e+02 & -4.69e-01 & -9.07e+00 & -2.19e-02 & & & 2.56e+01 & 3.38e-05 & & 0.984 & 5.00 & 274 & 68.32 & 295.36 + 40 & 1.73e+02 & 3.03e+01 & -6.04e-02 & 1.46e+01 & 3.33e-03 & -4.77e+00 & -5.47e-03 & 1.41e+01 & 8.31e-06 & & 0.901 & 2.90 & 281 & 9.05 & 53.33 + 41 & 8.41e+02 & 8.30e+01 & -1.76e-01 & -1.08e+02 & & -1.32e+01 & & 1.28e+01 & 1.58e-05 & 1.01e+01 & 0.872 & 3.24 & 282 & 60.27 & 110.59 + 42 & 3.85e+03 & 5.10e+02 & -1.86e+00 & 1.23e+03 & -1.13e-01 & 1.05e+02 & -1.53e-01 & 6.68e+01 & 1.94e-04 & -3.36e+01 & 0.990 & 13.00 & 271 & 208.55 & 892.36 + 43 & 9.45e+02 & 9.38e+01 & -2.98e-01 & 9.32e+01 & & & -1.65e-02 & 3.09e+01 & 2.66e-05 & & 0.960 & 6.21 & 272 & 36.16 & 235.48 + 44 & 2.88e+03 & 3.17e+02 & -1.03e+00 & 2.10e+02 & -3.75e-02 & & -5.87e-02 & 2.38e+01 & 1.03e-04 & 1.58e+01 & 0.981 & 5.88 & 268 & 77.20 & 359.05 + 45 & 2.07e+03 & 2.98e+02 & -7.48e-01 & 1.99e+02 & -3.47e-02 & & -3.36e-02 & 2.93e+01 & 7.03e-05 & & 0.981 & 5.26 & 268 & 49.94 & 302.09 + 46 & 7.37e+02 & 4.55e+01 & -1.28e-01 & -5.95e+01 & 8.18e-03 & -4.45e+00 & -1.22e-02 & -6.81e+00 & 1.16e-05 & 1.51e+01 & 0.981 & 3.39 & 273 & 37.12 & 182.41 + 47 & 1.05e+03 & 1.10e+02 & -1.98e-01 & -1.43e+02 & -7.00e-03 & & 2.38e-02 & 1.71e+01 & 5.31e-06 & & 0.977 & 3.24 & 270 & 17.62 & 141.32 + 48 & 1.75e+02 & -4.84e+01 & 5.51e-02 & -8.48e+01 & 2.02e-02 & & & & -4.46e-06 & 7.86e+00 & 0.966 & 3.37 & 275 & 64.88 & 158.64 + 49 & 1.37e+03 & 1.83e+02 & -6.10e-01 & 3.99e+02 & -1.86e-02 & 1.80e+01 & -3.15e-02 & 7.72e+00 & 5.66e-05 & -2.22e+01 & 0.980 & 7.33 & 271 & 127.06 & 406.41 + 50 & 8.61e+02 & 1.70e+02 & -4.43e-01 & 3.49e+02 & -3.51e-02 & 3.52e+01 & -2.57e-02 & 1.15e+01 & 4.13e-05 & -2.01e+01 & 0.979 & 5.83 & 269 & 80.91 & 288.34 + 51 & 8.41e+03 & 7.81e+02 & -3.58e+00 & 1.69e+03 & -1.86e-01 & 1.74e+02 & -2.65e-01 & 8.32e+01 & 3.66e-04 & & 0.991 & 20.48 & 279 & 274.01 & 1392.36 + 52 & 1.10e+03 & 7.03e+01 & -3.75e-01 & 8.20e+01 & & & -1.32e-02 & & 3.39e-05 & & 0.955 & 5.20 & 273 & 28.27 & 180.34 + 53 & 5.06e+02 & 2.42e+02 & -9.25e-02 & -5.18e+00 & -1.94e-02 & & & 7.31e+01 & 4.05e-06 & & 0.978 & 5.19 & 280 & 17.60 & 193.21 + 54 & 1.27e+03 & 2.03e+02 & -8.02e-01 & 7.14e+02 & -4.61e-02 & 4.92e+01 & -3.77e-02 & 1.31e+01 & 7.30e-05 & -5.24e+01 & 0.980 & 8.21 & 268 & 102.63 & 414.96 + 55 & 1.63e+03 & 2.50e+02 & -2.47e-01 & -2.93e+02 & -1.95e-02 & & 2.84e-02 & 3.21e+01 & 5.80e-06 & 1.50e+01 & 0.967 & 7.18 & 270 & 26.27 & 237.71 + 56 & 4.56e+03 & 7.69e+02 & -1.31e+00 & 1.32e-01 & -7.90e-02 & & & 1.04e+02 & 9.97e-05 & & 0.979 & 14.32 & 283 & 122.47 & 677.47 + 57 & 3.05e+03 & 1.95e+02 & -9.81e-01 & 1.11e+02 & -5.14e-02 & 5.26e+01 & -1.27e-02 & 1.73e+01 & 7.93e-05 & & 0.983 & 6.87 & 269 & 93.58 & 339.92 + 58 & 2.30e+03 & 1.26e+02 & -7.07e-01 & 1.07e+02 & -1.15e-02 & 2.25e+01 & & 5.85e+01 & 5.00e-05 & -1.23e+01 & 0.983 & 7.72 & 280 & 23.45 & 304.21 + 59 & 7.81e+03 & 8.83e+02 & -2.82e+00 & 6.88e+02 & -1.23e-01 & 2.92e+01 & -1.46e-01 & 5.02e+01 & 2.73e-04 & 2.62e+01 & 0.987 & 15.22 & 273 & 193.77 & 849.02 + 60 & 2.46e+03 & 3.62e+02 & -8.19e-01 & 1.90e+02 & -2.99e-02 & & & 4.67e+01 & 6.22e-05 & -2.16e+01 & 0.975 & 9.80 & 284 & 82.59 & 402.08 + 61 & 9.89e+03 & 5.59e+02 & -3.49e+00 & 9.09e+02 & -1.16e-01 & 1.42e+02 & -1.37e-01 & 6.19e+01 & 3.15e-04 & & 0.988 & 22.10 & 273 & 320.93 & 1298.99 + 62 & 5.90e+03 & 5.62e+02 & -2.24e+00 & 7.42e+02 & -6.33e-02 & 3.18e+01 & -1.20e-01 & 3.55e+01 & 2.18e-04 & & 0.978 & 18.75 & 278 & 220.46 & 984.53 + 63 & 1.52e+03 & 1.53e+02 & -6.41e-01 & 3.37e+02 & -3.86e-02 & 4.74e+01 & -2.00e-02 & & 5.46e-05 & -2.07e+01 & 0.964 & 9.44 & 276 & 102.89 & 359.97 + 64 & 1.34e+02 & 2.63e+00 & 3.44e-02 & -6.47e+01 & 7.56e-03 & & 1.12e-02 & 9.23e+00 & -8.53e-06 & & 0.870 & 4.85 & 277 & 12.81 & 77.37 + 65 & 2.17e+02 & 1.51e+01 & 1.06e-01 & -1.71e+02 & 1.11e-02 & & 2.53e-02 & 4.02e+01 & -1.86e-05 & & 0.855 & 8.58 & 274 & 47.36 & 155.93 + 66 & 1.96e+03 & 3.94e+02 & -6.06e-01 & 2.19e+02 & -2.02e-02 & & -3.18e-02 & 3.11e+01 & 5.30e-05 & & 0.969 & 14.72 & 281 & 152.05 & 574.88 + 67 & 7.35e+03 & 8.10e+02 & -2.64e+00 & 6.38e+02 & -9.40e-02 & & -9.95e-02 & & 2.42e-04 & & 0.980 & 18.04 & 277 & 191.54 & 962.50 + 68 & 7.73e+03 & 8.66e+02 & -2.91e+00 & 8.55e+02 & -1.54e-01 & 7.85e+01 & -1.37e-01 & 7.14e+01 & 2.76e-04 & & 0.984 & 17.69 & 277 & 177.46 & 845.72 + 69 & 2.80e+03 & 3.30e+02 & -6.29e-01 & -1.94e+02 & -2.18e-02 & & 3.22e-02 & 5.09e+01 & 3.18e-05 & & 0.984 & 8.41 & 265 & 49.75 & 419.83 + 70 & 4.39e+03 & 3.88e+02 & -1.87e+00 & 8.65e+02 & -9.39e-02 & 8.42e+01 & -1.38e-01 & 2.90e+01 & 1.94e-04 & & 0.991 & 9.61 & 269 & 192.69 & 735.89 + 71 & 1.41e+03 & 7.35e+01 & -6.60e-01 & 5.41e+02 & & 4.07e+01 & -4.16e-02 & 3.31e+01 & 6.10e-05 & -3.32e+01 & 0.985 & 9.27 & 269 & 129.70 & 536.39 + 72 & 6.93e+02 & 6.16e+01 & -1.99e-01 & 7.77e+01 & & 1.21e+01 & -1.33e-02 & 2.16e+01 & 1.74e-05 & & 0.976 & 5.31 & 274 & 47.53 & 236.68 + 73 & 2.68e+03 & 5.54e+02 & -8.35e-01 & 6.33e+01 & -7.21e-02 & & -1.09e-02 & 6.26e+01 & 6.71e-05 & & 0.987 & 5.89 & 270 & 20.75 & 300.08 + 74 & 1.19e+03 & 1.13e+02 & -4.96e-01 & 4.82e+02 & -1.03e-02 & 4.23e+01 & -7.88e-02 & 3.26e+01 & 5.82e-05 & & 0.990 & 8.03 & 276 & 202.99 & 604.88 + 75 & 1.24e+03 & 2.90e+02 & -1.33e-01 & -2.23e+02 & -1.82e-02 & -1.47e+01 & 1.99e-02 & 4.56e+01 & -3.53e-06 & 1.16e+01 & 0.982 & 5.58 & 280 & 17.49 & 237.20 + 76 & 2.32e+03 & 5.40e+02 & -6.80e-01 & 7.71e+01 & -4.92e-02 & -1.17e+01 & -1.32e-02 & 7.14e+01 & 5.41e-05 & & 0.990 & 6.64 & 282 & 57.41 & 405.60 + 77 & 2.91e+03 & 3.00e+02 & -9.33e-01 & 8.25e+01 & -2.74e-02 & -9.92e+00 & -5.32e-02 & 1.81e+01 & 9.31e-05 & 2.71e+01 & 0.983 & 5.63 & 271 & 80.82 & 346.14 + 78 & 1.88e+03 & 3.87e+02 & -6.33e-01 & 1.53e+02 & -3.50e-02 & -1.61e+01 & -4.72e-02 & 3.82e+01 & 6.64e-05 & 1.30e+01 & 0.972 & 6.41 & 279 & 87.22 & 317.56 + 79 & 4.64e+03 & 6.55e+02 & -1.63e+00 & 3.08e+02 & -7.55e-02 & -1.35e+01 & -5.24e-02 & 3.51e+01 & 1.48e-04 & & 0.983 & 8.92 & 274 & 90.97 & 506.80 + 80 & 2.28e+03 & 6.35e+02 & -6.00e-01 & -9.05e+01 & -6.72e-02 & -2.44e+01 & 1.44e-02 & 6.82e+01 & 3.91e-05 & & 0.984 & 5.91 & 274 & 4.81 & 262.53 + 81 & -1.78e+01 & 1.25e+01 & 4.93e-02 & -3.53e+00 & 5.45e-03 & & & & -6.20e-06 & & 0.944 & 2.97 & 275 & 10.24 & 73.26 + 82 & 2.33e+02 & 1.61e+02 & -3.68e-02 & 3.14e+01 & & -1.74e+01 & -7.89e-03 & 2.53e+01 & 3.65e-06 & & 0.973 & 3.82 & 273 & 23.46 & 145.70 + 83 & 1.68e+03 & 5.67e+02 & -4.07e-01 & -8.50e+01 & -5.96e-02 & -2.29e+01 & 1.31e-02 & 5.50e+01 & 2.38e-05 & & 0.958 & 8.34 & 282 & 17.03 & 237.91 + 84 & 4.80e+02 & 3.10e+02 & -2.64e-02 & -3.56e+01 & -9.49e-03 & -2.56e+01 & & 2.61e+01 & & & 0.969 & 7.01 & 287 & 70.26 & 272.69 + 85 & 9.04e+02 & 3.19e+02 & -3.12e-01 & 1.80e+02 & & -4.04e+01 & & 1.73e+01 & 2.16e-05 & -2.32e+01 & 0.943 & 10.38 & 289 & 50.34 & 288.82 + 86 & 1.60e+03 & 4.26e+02 & -4.31e-01 & -2.13e+01 & -2.43e-02 & -3.37e+01 & & 5.65e+01 & 3.27e-05 & & 0.949 & 9.23 & 288 & 27.81 & 269.08 + 87 & 8.09e+02 & 4.29e+02 & -1.23e-01 & 5.13e+00 & -2.48e-02 & -2.56e+01 & 2.35e-02 & 5.00e+01 & -5.56e-06 & -1.81e+01 & 0.977 & 7.58 & 279 & 59.74 & 312.23 + 88 & 2.96e+02 & 1.37e+02 & -1.74e-03 & -6.35e+01 & & -1.56e+01 & 9.61e-03 & 1.55e+01 & -5.55e-06 & & 0.956 & 4.04 & 290 & 13.13 & 117.75 + 89 & 1.43e+03 & 3.10e+02 & -2.88e-01 & -6.82e+01 & -1.04e-02 & -1.97e+01 & 1.02e-02 & 2.95e+01 & 1.36e-05 & & 0.986 & 5.70 & 275 & 62.96 & 297.99 + 90 & 1.17e+03 & 3.31e+02 & -3.68e-01 & 1.11e+02 & -3.58e-02 & & -1.81e-02 & 3.37e+01 & 3.18e-05 & & 0.989 & 4.28 & 275 & 48.27 & 262.34 + 91 & 9.60e+02 & 1.77e+02 & -2.73e-01 & 4.37e+01 & -6.48e-03 & -1.01e+01 & & 2.81e+01 & 1.90e-05 & -6.29e+00 & 0.989 & 3.00 & 266 & 22.05 & 157.34 + 92 & 1.40e+03 & 3.18e+02 & -6.53e-01 & 4.47e+02 & -6.81e-02 & 4.72e+01 & -5.77e-02 & 1.23e+01 & 6.69e-05 & -8.61e+00 & 0.991 & 4.93 & 263 & 83.85 & 347.46 + 93 & 1.85e+03 & 3.10e+02 & -8.84e-01 & 6.41e+02 & -6.29e-02 & 5.64e+01 & -5.19e-02 & 9.02e+00 & 8.04e-05 & -3.39e+01 & 0.990 & 7.30 & 271 & 113.74 & 474.63 + 94 & 1.11e+03 & 2.05e+02 & -5.11e-01 & 3.19e+02 & -3.74e-02 & 2.15e+01 & -3.49e-02 & 3.73e+00 & 5.06e-05 & -1.17e+01 & 0.986 & 3.77 & 265 & 62.06 & 265.81 + 95 & 4.04e+02 & 1.00e+02 & 4.59e-02 & -1.45e+02 & 1.62e-02 & -2.85e+01 & 2.16e-02 & 2.72e+01 & -1.56e-05 & & 0.956 & 4.74 & 270 & 12.75 & 129.88 + 96 & 1.36e+03 & -5.09e+01 & -9.19e-01 & 9.02e+02 & -2.51e-02 & 8.31e+01 & -9.26e-02 & -2.45e+01 & 9.79e-05 & -3.40e+01 & 0.988 & 8.72 & 266 & 113.80 & 514.74 + lrrrc calibration & 1 @xmath13 78 & 0.01 @xmath13 0.06 & 0.00 @xmath13 0.15 & 309 + marvels & @xmath897 @xmath13 110 & 0.00 @xmath13 0.07 & @xmath80.30 @xmath13 0.17 & 30 + elodie & 15 @xmath13 125 & 0.06 @xmath13 0.10 & 0.07 @xmath13 0.29 & 219 + calibration & @xmath83 @xmath13 80 & 0.01 @xmath13 0.06 & 0.00 @xmath13 0.15 & 309 + marvels & @xmath828 @xmath13 81 & 0.02 @xmath13 0.05 & @xmath80.07 @xmath13 0.15 & 30 +	studies of galactic chemical and dynamical evolution in the solar neighborhood depend on the availability of precise atmospheric parameters ( effective temperature , metallicity [ fe / h ] and surface gravity ) for solar - type stars . many large - scale spectroscopic surveys operate at low to moderate spectral resolution for efficiency in observing large samples , which makes the stellar characterization difficult due to the high degree of blending of spectral features . most surveys therefore employ spectral synthesis , which is a powerful technique , but relies heavily on the completeness and accuracy of atomic line databases and can yield possibly correlated atmospheric parameters . in this work , we use an alternative method based on spectral indices to determine the atmospheric parameters of a sample of nearby fgk dwarfs and subgiants observed by the marvels survey at moderate resolving power ( r @xmath0 12,000 ) . to avoid a time - consuming manual analysis , we have developed three codes to automatically normalize the observed spectra , measure the equivalent widths of the indices and , through the comparison of those with values calculated with pre - determined calibrations , determine the atmospheric parameters of the stars . the calibrations were derived using a sample of 309 stars with precise stellar parameters obtained from the analysis of high - resolution feros spectra , permitting the low - resolution equivalent widths to be directly related to the stellar parameters . a validation test of the method was conducted with a sample of 30 marvels targets that also have reliable atmospheric parameters derived from the high - resolution spectra and spectroscopic analysis based on excitation and ionization equilibria method . our approach was able to recover the parameters within 80 k for , 0.05 dex for [ fe / h ] and 0.15 dex for , values that are lower or equal to the typical external uncertainties found between different high - resolution analyzes . an additional test was performed with a subsample of 138 stars from the elodie stellar library and the literature atmospheric parameters were recovered within 125 k for , 0.10 dex for [ fe / h ] and 0.29 dex for . these precisions are consistent or better than those provided by the pipelines of surveys operating with similar resolutions . these results show that the spectral indices are a competitive tool to characterize stars with the intermediate resolution spectra .
recent analyses of extended rotation curves ( rcs ) of late - type spiral galaxies ( persic , salucci & stel 1996 ) have confirmed that in spirals of all luminosities a substantial dark matter ( dm ) component is detectable already in the optical region . the effect is stronger at lower luminosities : the dark - to - visible mass ratio at the optical radius @xmath0 is the radius encompassing the @xmath10 of the total integrated light . @xmath11 is the scale - length of the exponential surface brightness distribution @xmath12 . for a freeman ( 1970 ) disk , @xmath0 corresponds to the de vaucouleurs 25 @xmath13- photometric radius . ] scales with luminosity @xmath14 . for early - type spirals the status of our knowledge is different . the rcs presently available for these objects are fragmentary ( in particular in the nuclear regions ) , and only extend to @xmath15 ( see rubin et al . consequently , detailed mass decomposition have so far not been possible for these systems . in particular , it is not known whether dark halos are unambiguously present also in early - type spirals . it may be conjectured that for a given @xmath16 the dm fraction within the optical size is smaller in early than in late - type spirals : this , because in an early spiral the conspicuous stellar bulge , with @xmath17 , can supply a mass compact enough to make the rotation velocity higher than ( see rubin et al . 1985 ) , and the velocity profile different from , that of a late spiral of similar luminosity . in this case , the derivation of the halo parameters would be more uncertain for early than for late types : in fact , at small radii not two mass components ( disk + halo , like in sc - sd galaxies ) , but three mass components ( bulge + disk + halo ) will have locally similar ( solid - body like ) behaviors . so for non - extended rc data the mass solution of an sa galaxy would be degenerate even within the maximum - disk solution . in this paper we present the velocity and velocity - dispersion profiles of the stars and the ionized gas , measured along the major axis , for six early - type spirals . the six selected galaxies ( table 1 ) were already known to show emission lines and their photometric properties were known . of these , 5 had already been observed spectroscopically by rubin et al . ( 1985 ) , who obtained the rcs of the ionized gas , and photometrically by kent ( 1988 ) . ngc 2179 was the only galaxy in our sample still lacking spectroscopical and photometrical observations . to the originally observed sample belonged the early - type spiral too . its stellar and the gaseous kinematics , found to exhibit a star vs. star counterrotation , is presented and discussed by bertola et al . three - component models ( bulge + disk + halo ) based on observed photometry and kinematics are obtained for two galaxies of the sample : ngc 2179 and ngc 2775 . the spectroscopic observations of our sample galaxies were carried out at the eso 1.52-m spectroscopic telescope at la silla on february 15 - 19 , 1994 . the telescope was equipped with the boller & chivens spectrograph . the no . 26 grating with 1200 @xmath18 was used in the first order in combination with a @xmath19 slit . it yielded a wavelength coverage of 1990 between about 5200 and about 7190 with a reciprocal dispersion of 64.80 mm@xmath7 . the instrumental resolution was derived measuring after calibration the fwhm of 22 individual emission lines distributed all over the spectral range in the 10 central rows of a comparison spectrum . we checked that the measured fwhm s did not depend on wavelength , and we found a fwhm mean value of 2.34 . this corresponds to @xmath20 ( i.e. , @xmath21 at 5800 and @xmath22 at 6400 ) . the adopted detector was the no . 24 2048@xmath232048 ford ccd , which has a @xmath24m@xmath25 pixel size . after an on - chip binning of 3 pixels along the spatial direction , each pixel of the frame corresponds to 0.97 @xmath26 . the long - slit spectra of all the galaxies were taken along their optical major axes . at the beginning of each exposure the galaxy was centered on the slit using the guiding camera . repeated exposures ( typically of 3600 s each ) did ensure several hours of effective integration without storing up too many cosmic rays . some long - slit spectra of 8 late - g or early - k giant stars , obtained with the same instrumental setup , served as templates in measuring the stellar kinematics . their spectral classes range from g8iii to k4iii ( hoffleit & jaschek 1982 ) . the typical value of the seeing fwhm during the observing nights , measured by the la silla differential image motion monitor ( dimm ) , was @xmath27 . comparison helium - argon lamp exposures were taken before and after every object exposure . the logs of the spectroscopic observations of galaxies and template stars are reported in tables 2 and 3 , respectively . llcc & & & + & & & + & & & + & 15 feb 1994 & 01 24 & 3600 + ngc 2179 & 15 feb 1994 & 02 32 & 3600 + ngc 2179 & 16 feb 1994 & 01 17 & 5400 + ngc 2179 & 17 feb 1994 & 01 41 & 3600 + ngc 2179 & 18 feb 1994 & 00 26 & 3600 + ngc 2179 & 18 feb 1994 & 01 35 & 3600 + ngc 2179 & 19 feb 1994 & 00 19 & 3600 + ngc 2179 & 19 feb 1994 & 01 27 & 3600 + & 15 feb 1994 & 04 10 & 3600 + ngc 2775 & 15 feb 1994 & 05 17 & 3600 + ngc 2775 & 16 feb 1994 & 03 56 & 5400 + ngc 2775 & 17 feb 1994 & 02 47 & 3600 + ngc 2775 & 17 feb 1994 & 03 53 & 3600 + ngc 2775 & 18 feb 1994 & 02 41 & 3600 + & 16 feb 1994 & 04 39 & 5400 + ngc 3281 & 16 feb 1994 & 06 14 & 3600 + ngc 3281 & 17 feb 1994 & 05 02 & 3600 + ngc 3281 & 18 feb 1994 & 03 50 & 3600 + ngc 3281 & 19 feb 1994 & 02 34 & 3600 + ngc 3281 & 19 feb 1994 & 03 40 & 3600 + & 17 feb 1994 & 06 16 & 3600 + ic 724 & 17 feb 1994 & 07 21 & 3600 + & 16 feb 1994 & 07 22 & 3600 + ngc 4698 & 16 feb 1994 & 08 28 & 3600 + ngc 4698 & 18 feb 1994 & 07 17 & 3600 + ngc 4698 & 19 feb 1994 & 07 42 & 6000 + & 15 feb 1994 & 07 02 & 3600 + ngc 4845 & 15 feb 1994 & 08 08 & 3600 + ngc 4845 & 17 feb 1994 & 05 02 & 3600 + ngc 4845 & 18 feb 1994 & 08 25 & 3600 + notes cols.(2 - 3 ) : date and time of start of exposure . col.(4 ) : exposure time . [ tab : log_g ] lllcc & & & & + & & & & + & & & & + & g8iii & 15 feb 1994 & 00 53 & 2 + hr 2035 & g8iii & 17 feb 1994 & 00 04 & 2 + hr 2035 & g8iii & 17 feb 1994 & 00 07 & 3 + hr 2035 & g8iii & 17 feb 1994 & 00 10 & 4 + hr 2035 & g8iii & 17 feb 1994 & 00 16 & 3 + hr 2035 & g8iii & 19 feb 1994 & 00 07 & 2 + hr 2035 & g8iii & 19 feb 1994 & 00 09 & 3 + & k1iii & 18 feb 1994 & 00 08 & 2 + hr 2429 & k1iii & 18 feb 1994 & 00 11 & 2 + hr 2429 & k1iii & 18 feb 1994 & 00 14 & 3 + & k1iii & 16 feb 1994 & 00 25 & 3 + & k4iii & 16 feb 1994 & 00 35 & 4 + hr 2503 & k4iii & 16 feb 1994 & 00 38 & 4 + hr 2503 & k4iii & 16 feb 1994 & 00 41 & 4 + hr 2503 & k4iii & 16 feb 1994 & 00 46 & 4 + hr 2503 & k4iii & 16 feb 1994 & 00 49 & 8 + & k4iii & 15 feb 1994 & 03 52 & 4 + & k0iii & 19 feb 1994 & 09 29 & 10 + hr 5100 & k0iii & 19 feb 1994 & 09 32 & 7 + & k0.5iii & 16 feb 1994 & 09 33 & 4 + hr 5196 & k0.5iii & 16 feb 1994 & 09 38 & 15 + hr 5196 & k0.5iii & 16 feb 1994 & 09 43 & 10 + & k3iii & 15 feb 1994 & 09 14 & 4 + hr 5315 & k3iii & 15 feb 1994 & 09 17 & 3 + hr 5315 & k3iii & 15 feb 1994 & 09 21 & 3 + & k0.5iii & 15 feb 1994 & 09 29 & 3 + hr 5601 & k0.5iii & 15 feb 1994 & 09 32 & 4 + hr 5601 & k0.5iii & 15 feb 1994 & 09 35 & 4 + hr 5601 & k0.5iii & 15 feb 1994 & 09 39 & 10 + notes col.(2 ) : spectral class of the template star from the bright star catalogue ( hoffleit & jaschek 1982 ) . cols.(3 - 4 ) : date and time of start of exposure . col.(5 ) : exposure time . using standard midas routines , all the spectra were bias subtracted , flat - field corrected by quartz lamp exposures , and cleaned from cosmic rays . cosmic rays were identified by comparing the counts in each pixel with the local mean and standard deviation , and then corrected by substituting a suitable value . a small misalignment was present between the ccd and the slit . we measured a difference of @xmath28 pixel between the positions of the center of the stellar continuum near the blue and red edge of the spectra . when measuring the stellar kinematics , the tilt had to be removed . this was done by rotating the spectra by a suitable angle ( @xmath29 ) before the wavelength calibration . we noticed however that the sharp line profile of the emission lines was spoiled by the rotating algorithm . for this reason no rotation was applied when measuring the ionized - gas kinematics . the wavelength calibration was done using the midas package xlong . we determined the velocity error possibly introduced by the calibration measuring the ` velocity curve ' of a sample of 24 oh night - sky emission lines distributed all over the spectral range . the velocity did not show any significant dependence on radius , indicating that the wavelength rebinning had been done proper - ely . we found a mean deviation from the predicted wavelengths ( osterbrock et al . 1996 ) of @xmath30 . after calibration , the different spectra obtained for a given galaxy were co - added using their stellar - continuum centers as reference . for each spectrum the center was assumed as the center of the gaussian fitting the mean radial profile of the stellar continuum . the contribution of the sky was determined from the edges of the resulting galaxy frames and then subtracted . the ionized - gas velocities ( @xmath31 ) and velocity dispersions ( @xmath32 ) were measured by means of the midas package alice . we measured the lines ( @xmath336548.03 , 6583.41 ) , the line ( @xmath346562.82 ) , and the lines ( @xmath336716.47 , 6730.85 ) , where they were clearly detected . the position , the fwhm , and the uncalibrated flux @xmath35 of each emission line were determined by interactively fitting one gaussian to each line plus a polynomial to its local continuum . the center wavelength of the fitting gaussian was converted into velocity in the optical convention @xmath36 ; then the standard heliocentric correction was applied . the gaussian fwhm was corrected for the instrumental fwhm , and then converted into the velocity dispersion @xmath37 . in the regions where the intensity of the emission lines was low , we binned adjacent spectral rows in order to improve the signal - to - noise ratio , @xmath38 , of the lines . we expressed the variation of the r.m.s . velocity error @xmath39 as a function of the relevant line @xmath38 ratio . in order to find the expression for @xmath40 , we selected the same 16 night - sky emission lines in the spectra of ngc 2775 , ngc 3281 , ic 724 and ngc 4845 . such night - sky emissions were chosen to have different intensities and different wavelengths between 6450 and 6680 (i.e . the wavelength range of the observed emission lines of the ionized gas ) in the four spectra . we derived the sky spectra by averaging several rows along the spatial direction in a galaxy - light free region . using the above package , we interactively fitted one gaussian emission plus a polynomial continuum to each selected sky line and its local continuum . we derived the flux @xmath35 and the fwhm of the sample lines , taking the ratio @xmath41 as the signal @xmath42 . for each galaxy spectrum the noise @xmath43 was defined as the r.m.s . of the counts measured in regions of the frame where the contributions of both the galaxy and the sky lines were negligible . the resulting @xmath38 range was large ( @xmath44 ) . for each sample emission line we then measured , by means of an automatic procedure , the night - sky ` velocity curve ' along the full slit extension . the wavelengths of the emissions were evaluated with gaussian fits and then converted to velocities . the radial profiles of the sky - line velocities were then fitted by quadratic polynomials . we assumed the r.m.s . of the fit to each ` velocity curve ' to be the @xmath45 velocity error . [ fig : snratio ] shows the good agreement between the distributions of the @xmath46 measurements taken in the 4 different spectra . in log - log scale , the @xmath47 relation is well represented by a straight line , that corresponds to : @xmath48 ( least - squares fit ) . this result agrees with keel s ( 1996 ) relation @xmath49 , based on numerical simulations . once the relevant @xmath38 ratio of the emission had been derived , we obtained @xmath39 for each velocity measurement of the ionized - gas component by means of eq . [ eq : sn ] . the gas velocities derived independently from different emission lines are in mutual agreement within their errors @xmath39 . the ionized - gas velocities and velocity dispersions from ( @xmath336548.03 , 6583.41 ) , ( @xmath336716.47 , 6730.85 ) , and are reported in : tables 4 7 for ngc 2179 ; tables 10 14 for ngc 2775 ; tables 17 21 for ngc 3281 ; tables 24 26 for ic 724 ; tables 29 33 for ngc 4698 ; and tables 36 40 for ngc 4845 . each table reports the galactocentric distance @xmath50 in arcsec ( col . 1 ) , the observed heliocentric velocity @xmath51 and its error @xmath39 in ( col . 2 ) , the velocity dispersion @xmath37 in ( col . 3 ) , the number @xmath52 of spectrum rows binned along the spatial direction ( col . 4 ) , and the signal - to - noise ratio @xmath38 of the emission line ( col . 5 ) . the , and kinematics of the sample objects are plotted in figs . [ fig : gas_a ] [ fig : gas_c ] . the final ionized - gas kinematics is obtained by averaging , at each radius , the gas velocities and velocity dispersions derived independently from the different emission lines . the gas velocity ( @xmath31 ) and velocity error ( @xmath53 ) are respectively the @xmath54-weighted mean velocity and its uncertainty . the gas velocity dispersion ( @xmath32 ) and velocity - dispersion error ( @xmath55 ) are the mean velocity dispersion and its uncertainty . ( no error is given when only one velocity dispersion measurement is available . ) the kinematics of the ionized gas is reported in : table 8 for ngc 2179 ; table 15 for ngc 2775 ; table 22 for ngc 3281 ; table 27 for ic 724 ; table 34 for ngc 4698 ; and table 41 for ngc 4845 . each table reports the galactocentric distance @xmath50 in arcsec ( col . 1 ) , the mean heliocentric velocity @xmath31 and its error @xmath53 in ( col . 2 ) , the mean velocity dispersion @xmath32 and its error @xmath55 in ( col . 3 ) . the ionized - gas velocity and velocity - dispersion profiles are plotted in fig . [ fig : kin ] for all our galaxies . the stellar velocities ( @xmath56 ) and velocity dispersions ( @xmath57 ) of the sample galaxies were measured from the absorption lines in the wavelength range between about 5200 and 6200 . we used an interactive version of the fourier quotient method ( sargent et al . 1977 ) as applied by bertola et al . the k0iii star hr 5100 was taken as template : it has a radial velocity of @xmath58 ( wilson 1953 ) and a rotational velocity of @xmath59 ( bernacca & perinotto 1970 ) . the spectra of the galaxies and the template star were rebinned to a logarithmic wavelength scale , continuum subtracted , and masked at their edges by means of a cosine bell function of @xmath60 length . at each radius the galaxy spectrum was assumed to be the convolution of the template spectrum with a gaussian broadening function characterized by the parameters @xmath61 , @xmath56 and @xmath57 . they respectively represent the line strength of the galaxy spectrum relative to the template s , and the line - of - sight stellar velocity and velocity dispersion . the parameters of the broadening function , and consequently the stellar kinematics , were obtained by a least - squares fitting in the fourier space of the broadened template spectrum to the galaxy spectrum in the wavenumber range @xmath62=\left[5,440\right]$ ] . in this way we rejected the low - frequency trends ( corresponding to @xmath63 ) due to the residuals of continuum subtraction and the high - frequency noise ( corresponding to @xmath64 ) due to the instrumental resolution . ( the wavenumber range is important in particular in the fourier fitting of lines with non - gaussian profiles , see van der marel & franx 1993 and cinzano & van der marel 1994 ) . in deriving the above kinematical properties , the regions @xmath65 and @xmath66 were masked because of contamination from bad subtraction of the night - sky emission lines of ( @xmath67 ) and ( @xmath68 ) . the measured stellar kinematics is reported in : table 9 for ngc 2179 ; table 16 for ngc 2775 ; table 23 for ngc 3281 ; table 28 for ic 724 ; table 35 for ngc 4698 ; and table 42 for ngc 4845 . each table reports the galactocentric distance @xmath50 in arcsec ( col . 1 ) , the heliocentric velocity @xmath56 and its error @xmath69 in ( col . 2 ) , the velocity dispersion @xmath57 and its error @xmath70 in ( col . 3 ) . the stellar velocity and velocity - dispersion profiles are plotted in fig . [ fig : kin ] . the observation in the cousin @xmath71band of ngc 2179 was performed on march 11 , 1997 at the 1.83-m vatican advanced technology telescope ( vatt ) at mt . graham international observatory . a back - illuminated 2048@xmath232048 loral ccd with @xmath72@xmath73m@xmath25 pixels was used as detector at the aplanatic gregorian focus , f/9 . it yielded a field of view of @xmath74 with an image scale of @xmath75 pixel@xmath7 after a @xmath76 pixel binning . the gain and the readout noise were 1.4 e@xmath77 adu@xmath7 and 6.5 e@xmath77 respectively . we obtained 3 images of 120 s with the @xmath78 3.48-inch square filter . the data reduction was carried out using standard iraf routines . the images were bias subtracted and then flat - field corrected . they were shifted and aligned to an accuracy of a few hundredths of a pixel using field stars as reference . after checking that the point spread functions ( psfs ) in the images were comparable , they were averaged to obtain a single @xmath78 image . the cosmic rays were identified and removed during the averaging routine . a gaussian fit to the intensity profile of field stars in the resulting image allowed us to estimate a seeing psf fwhm of @xmath79 . the sky subtraction and the elliptical fitting to the galaxy isophotes were performed by means of the astronomical images analysis package ( aiap : fasano 1990 ) . the sky level was determined by a polynomial fit to surface brightness in frame regions not contaminated by galaxy light ; then it was subtracted out from the total signal . the isophote fitting was performed masking the frame s bad columns and the bright field stars . we then obtained surface brightness , ellipticity , major - axis position angle , and the @xmath80 fourier coefficient of the isophote s deviations from elliptical as a function of radius along the major axis . no photometric standards were observed . thus the absolute calibration was made using the photometric quantities edited by lauberts & valentijn ( 1989 ) in the same band . we set the surface brightness of an isophote with semi - major axis of @xmath81 to the value @xmath82 @xmath78- . the resulting kinematics of all our galaxies are shown in fig . [ fig : kin ] . the plotted velocities are as observed ( no inclination correction is applied ) . in the following we briefly discuss each individual object . at each radius , @xmath83 ( @xmath84 ) are the observed rotation velocities of the stars and the ionized gas , respectively . + * ngc 2179 * the gas and stellar kinematics respectively extend to @xmath85 ( @xmath86 kpc ) and @xmath87 ( @xmath88 kpc ) on either side of the nucleus . outwards of @xmath87 , the stellar and the gas radial velocities are comparable . + _ stars . _ in the inner @xmath89 ( @xmath90 kpc ) , @xmath83 increases to @xmath91 . at the center @xmath92 ; away from the nucleus it remains high ( @xmath93 ) , and possibly even rises . _ @xmath94 has a steeper gradient than @xmath83 , reaching a value of @xmath95 at @xmath96 ( @xmath97 kpc ) . @xmath32 is strongly peaked ( @xmath98 ) at the center ; at @xmath99 it drops rapidly to @xmath100 . + a circumnuclear keplerian disk of ionized gas has been recently discovered in the center of ngc 2179 by bertola et al . ( 1998a ) by means of optical ground - based observations . by modeling the motion of the gaseous disk they inferred the presence of a central mass concentration of @xmath101 . + * ngc 2775 * the gas and stellar kinematics is measured out to @xmath102 ( @xmath103 kpc ) from the center . + _ stars . _ @xmath83 increases almost linearly with radius , up to about 130 at @xmath104 ( @xmath90 kpc ) ; for @xmath105 it remains approximately constant ; further out it increases to 185 and then flattens out . at @xmath106 , @xmath107 ; farther out it declines to @xmath108 and @xmath91 in the se and nw side , respectively . _ the gas behaves differently in the two regions @xmath109 ( @xmath110 kpc ) and @xmath111 . the line ( the only emission line detected in both regions , see fig . [ fig : gas_a ] ) shows the presence of two kinematically distinct gas components , named component ( i ) and ( ii ) . component ( i ) rotates with the same velocity as the stars but with a lower velocity dispersion ; component ( ii ) rotates faster than the stars for @xmath112 ( 1.1 kpc ) . both components show up simultaneously in the spectrum only at @xmath113 , where a double peak in the emission line is clearly detected . @xmath32 peaks ( 160 ) at the center and rapidly drops to @xmath114 off center . + * ngc 3281 * the stellar kinematics extends to @xmath115 ( @xmath116 kpc ) and @xmath117 ( 12 kpc ) in the se and the nw side , respectively . the gas kinematics can only be measured within @xmath85 ( 10 kpc ) on each side of the nucleus . + _ stars . _ the stars exhibit a rather shallow rotation gradient : at @xmath118 ( @xmath30 kpc ) , @xmath119 . at the center @xmath120 ; off center it decreases to , respectively , @xmath121 in the se side and @xmath122 in the nw side . _ @xmath94 has a steep gradient , reaching 150 at @xmath89 and then 200 at @xmath123 ( 4 kpc ) . at the center @xmath124160 , while at @xmath111 it falls to @xmath12550 ; @xmath126 for @xmath127 . + * ic 724 * the stellar kinematics is observed out to @xmath117 ( @xmath128 kpc ) and @xmath87 ( @xmath129 kpc ) in the sw and ne sides , respectively . the ionized - gas kinematics extends to @xmath117 on each side of the nucleus . for @xmath111 ( @xmath130 kpc ) , the gas and stellar kinematics are similar . for @xmath131 , @xmath32 is centrally peaked at @xmath132 , remaining lower than @xmath57 . + _ stars . _ @xmath83 increases linearly up to @xmath133 in the inner @xmath134 ( @xmath135 kpc ) , followed by a drop to @xmath136 between @xmath134 and @xmath137 on both sides of the nucleus ; further out it rises to @xmath138 at @xmath123 , and then remains constant . at the center @xmath139 , then off the nucleus it decreases to @xmath122 _ @xmath94 has a steeper gradient than @xmath83 , peaking at 300 at about @xmath134 ; then it decreases , becoming @xmath140 at @xmath123 . the shows steeper central rc gradient and lower velocity dispersion than and ( see fig . [ fig : gas_b ] ) . this feature is probably due to the lower @xmath83 : the absorption does not have the same central wavelength as the emission , and hence it shifts the resulting peak toward higher rotation velocities . + * ngc 4698 * the stellar and ionized - gas kinematics are measured out to @xmath141 ( @xmath142 kpc ) and @xmath143 ( @xmath144 kpc ) on each side of the nucleus , respectively . + _ stars . _ in the innermost @xmath89 ( @xmath145 kpc ) the stars have zero rotation ; at outer radii , @xmath83 is less steep than @xmath94 ; only for @xmath146 ( 2.3 kpc ) are @xmath147 and @xmath148 . the profile of @xmath57 is radially asymmetric : in the se side it shows a maximum of @xmath98 at @xmath149 ( @xmath150 kpc ) , then it decreases outwards to @xmath122 and 30 at @xmath151 ( @xmath152 kpc ) in the se and nw sides , respectively . the measured @xmath153 is explained by bertola et al . ( 1998b ) as due to the presence of an orthogonal - rotating bulge . _ @xmath94 increases to @xmath125130 in the inner @xmath154 ( @xmath155 kpc ) ; then it increases more gradually reaching @xmath125200 at @xmath117 ( @xmath156 kpc ) , and stays approximately constant farther out . in the inner @xmath157 ( @xmath158 kpc , roughly coinciding with the absorption lines region ) @xmath32 has a @xmath159 plateau , while at larger radii it drops to @xmath160 . + * ngc 4845 * the stellar and ionized - gas kinematics are measured out to @xmath161 ( @xmath142 kpc ) in the sw side , and out to @xmath115 ( @xmath162 kpc ) in the ne side . + _ stars . _ @xmath83 has a shallower gradient than @xmath94 : it reaches 60 at @xmath134 ( 0.5 kpc ) , and further out it increases slowly , reaching the @xmath94 at @xmath163 ( 3.8 kpc ) . the velocity dispersion is constant , @xmath164 ( in the sw side it drops to 30 for @xmath165 ) . _ @xmath94 reaches @xmath125180 at @xmath166 ( 0.9 kpc ) , to decrease and remain constant at 150 farther out . for @xmath167 ( 0.6 kpc ) @xmath168 , then it rapidly falls to @xmath169 and @xmath108 in the sw and ne sides , respectively ; farther out the behavior of @xmath32 is more uncertain , due to a considerable scatter of the measurements from different lines : along the nw side @xmath170 , while along the se side @xmath32 slowly decreases to @xmath114 . + the triaxiality of the bulge of ngc 4845 has been detected by bertola , rubin & zeilinger ( 1989 ) and discussed by gerhard , vietri & kent ( 1989 ) . the rcs and velocity - dispersion profiles of both the ionized gas and the stars in our sa galaxies show a rich diversity of kinematical properties . @xmath83 has shallower gradient than @xmath94 at the center , while @xmath171 at the last measured radius , in all our sample galaxies . for ngc 2179 , ngc 2775 , ngc 3593 , ngc 4698 and ngc 4845 , the gas rcs remain flat after a monotonic rise to a maximum [ whose observed values range between @xmath172 ( as for ngc 3593 ) and @xmath173 ( as for ngc 2179 , ngc 2775 ) ] , or rise monotonically to the farthest observed radius ( as for ngc 3281 and ic 724 : in the latter after an initial peak at @xmath174 ) . stellar counterrotation and orthogonal rotation has been found in ngc 3593 ( bertola et al . 1996 ) and ngc 4698 ( bertola et al . 1998 ) respectively . the observed @xmath57 exceeds 100 for several kpc in the innermost regions , peaking at values ranging between 130 ( as in ngc 4698 ) and 210 ( as in ic 724 ) ; the only exception is ngc 4845 with @xmath175 at all observed radii . there are sample galaxies whose @xmath32 is low at all radii , reaching a central maximum of @xmath176 ( ngc 4698 , ngc 4845 ) or remaining flattish at @xmath122 ( ngc 3593 ) ; and others where @xmath32 increases to @xmath177 either at the very center ( as in ngc 2179 , ngc 2775 ) or over an extended radial range around the center ( as in ngc 3281 , ic 724 ) . previous authors ( fillmore , boroson & dressler 1986 ; kent 1988 ; kormendy & westpfahl 1989 ) noticed that in the bulge of early - type spirals @xmath94 falls below the predicted circular velocity . such ` slowly rising ' gas rcs are explained by bertola et al . ( 1995 ) with the argument that random ( non - circular ) motions are crucial for the dynamical support of the ionized gas : in some galaxies of their s0 sample they measured @xmath178 over an extended range of radii . we do observe the same phenomenon in some of our early - type spirals ( see fig . 5 ) : this fact prevents us to adopt , for early - type disk galaxies , the inner portion of @xmath179 as the circular velocity on which to perform the mass decomposition . when the high values for the velocity dispersion of the gas are measured only in the very central parts ( as in ngc 2179 ) we can not exclude that this is an effect of rotational broadening due to the seeing smearing of the steep velocity gradient . at larger radii where @xmath180 , the ionized gas can be considered a tracer of the actual circular velocity . but the limited extension , ( @xmath181 ( see tab . 1 ) , of our @xmath179 makes the derivation of the halo parameters of early - type spirals more uncertain than for later types . in fact , on one hand we lack data at very large radii where only disk and halo affect the circular velocity , on the other hand at small radii ( where we can not consider the gas in circular motion ) not two ( like for sc - sd galaxies ) , but three mass components will have locally similar behaviors ( solid - body like ) . so , in absence of extended and complete rcs , an sa mass solution would be degenerate . we therefore have to model the stellar kinematics to determine the galaxy s total gravitational potential . then we need to check the derived mass decomposition by comparing the circular velocity , inferred from the model , with @xmath94 at large galactocentric radii . this is necessary to minimize uncertainties on the mass structure obtained from the stellar kinematics . in fact , the uncertain orbital structure of the spheroidal component , consistent with the observed kinematics leads to a degeneracy between velocity anisotropy and mass distribution , which can be solved only through the knowledge of the line - of - sight velocity distribution profiles ( gerhard 1993 ) . two galaxies of the observed sample , namely ngc 2179 and ngc 2775 , are particularly suited to be studied with the three - component mass models , based on stellar photometry and kinematics , at our disposal . they were chosen for their nearly axisymmetric stellar pattern and to not contain kinematically decoupled ( as found ngc 3593 and ngc 4698 ) or triaxial ( as found in ngc 4845 ) stellar components . we apply the jeans modeling technique introduced by binney , davies & illingworth ( 1990 ) , developed by van der marel , binney & davies ( 1990 ) and van der marel ( 1991 ) , and extended to two - component galaxies by cinzano & van der marel ( 1994 ) and to galaxies with a dm halo by cinzano ( 1995 ) . ( for a detailed description of the model and its assumptions , see the above references . ) the galaxy is assumed to be axisymmetric . its mass structure results from the contributions of : _ ( i ) _ a spheroidal component ; _ ( ii ) _ an infinitesimally thin exponential disk ; and _ ( iii ) _ a spherical pseudo - isothermal dark halo with density distribution @xmath182 $ ] . the mass contribution of the ionized gas is assumed to be negligible at all radii . the spheroidal and disk components are supposed to have constant @xmath183 ratios . the total potential is the sum of the ( numerically derived ) potential of the spheroid plus the ( analytical ) potentials of the disk and the halo . the stellar distribution function @xmath184 is assumed to depend only on two integral of motion [ i.e. , @xmath185 . in these hypotheses the jeans equations for hydrostatic equilibrium form a closed set that , once solved in the total potential , yields the dynamical quantities to be compared with the observed kinematics , once projected onto the sky plane . to obtain the potentials of the bulge and the disk , we proceed through several steps . _ ( a ) _ first , the bulge surface brightness is derived from the total one by subtracting the disk . then , it is deprojected by means of lucy s algorithm to yield the 3-d luminosity density which , via the @xmath183 ratio , gives the 3-d mass density of the bulge . finally , solving poisson s equation through multipole expansion , we derive the bulge potential ( binney et al . _ ( b ) _ the exponential disk parameters ( scale length @xmath186 , central surface brightness @xmath187 , and inclination @xmath188 ) are chosen according to the best - fit photometric decomposition . if @xmath189 , @xmath190 and @xmath191 are the disk parameters resulting from the photometric decomposition , the best - fit model to the observed stellar kinematics was obtained considering exponential disks with @xmath192 , @xmath193 @xmath194 , and @xmath195 . these parameters determine the surface brightness of the disk . through the disk @xmath183 ratio we obtain the surface mass density of the disk , and then its potential ( binney & tremaine 1987 ) . we first solved the jeans equations only for both the bulge and disk components in their total potential , to give in every point of the galaxy the velocity dispersions onto the meridional plane @xmath196 and the mean azimuthal squared velocities @xmath197 . to disentangle the respective contributions of the azimuthal velocity dispersion @xmath198 and the mean stellar motion @xmath199 to @xmath197 , for the bulge we made the same hypotheses of binney et al . ( 1990 ) while for the disk we followed cinzano & van der marel ( 1994 ) respectively . part of the second azimuthal velocity moment @xmath200 in the bulge is assigned to the streaming velocity @xmath201 as in satoh ( 1980 ) . the azimuthal velocity dispersion @xmath202 in the disk is assumed to be related to @xmath203 ( which is assumed in turn to have an exponential fall - off with central value @xmath204 and scale - length @xmath205 ) according to the epicyclic theory ( cfr . binney & tremaine 1987 ) . in the framework of cinzano & van der marel ( 1994 ) , we have to take into account the effects of seeing , of finite slit - width and pixel - size in data acquisition , and of fourier filtering in data reduction ( notably the wavenumber range , as discussed in sect . 2.1 . ) , in order to compare the sky - projected model predictions with the observed stellar kinematics . we interpret the discrepancy between the model s circular velocity and the observed gas rotation in the outer regions as due to the presence of a dm halo . in this case , the jeans equations have to be solved again , taking the halo into account , too . by introducing the dm halo , the number of free parameters of the model increases to ten . they are : @xmath206 , the local rotation anisotropy parameter of the bulge ; the @xmath183 ratios of the bulge and the disk ; the disk central surface brightness , scale length and inclination ; the central value and scale length of the disk s second radial velocity moment ; and the halo s central mass density and core radius . to reduce the number of free parameters , in the following we consider only three - component models having same best - fit parameters as no - halo models except for the bulge and disk @xmath183 ratios . we choose the fit parameters in order to simultaneously reproduce the stellar kinematics at all radii as well as @xmath94 at large radii the modeling technique described above derives the 3-d distribution of the luminous mass from the 3-d luminosity distribution inferred from the observed surface photometry . for this reason , in the central regions we take into account the seeing effects on the measured photometrical quantities ( surface brightness , ellipticity and @xmath207 deviation profiles ) . we derive for ngc 2179 and ngc 2775 the seeing cutoffs @xmath208 and @xmath209 , defined by peletier et al . ( 1990 ) as the radii beyond which the seeing - induced error on the profile is lower than , respectively , 0.05 in surface brightness and 0.02 in ellipticity . they have expressed @xmath208 and @xmath209 for a de vaucouleurs profile as a function of the seeing fwhm , the effective radius @xmath210 and the ellipticity @xmath211 . for the bulge , we obtained @xmath210 and @xmath211 , and the corresponding seeing cutoffs @xmath208 and @xmath209 , following an iterative procedure . we started by performing a standard bulge - disk decomposition with a parametric fit ( e.g. kent 1985 ) : we decomposed the observed surface - brightness profile on both the major and the minor axis as the sum of a de vaucouleurs bulge of surface - brightness profile @xmath212\,,\ ] ] plus an exponential disk of surface - brightness profile @xmath213 we assumed the minor - axis profiles of each component to be the same as the major - axis profiles , with values scaled by a factor @xmath214 . a least - squares fit of the model to the photometric data provided @xmath210 , @xmath215 and @xmath211 of the bulge , @xmath187 , @xmath186 of the disk , and the galaxy inclination @xmath188 . the values of @xmath210 and @xmath211 were used as a starting input to derive @xmath208 and @xmath209 . following van der marel ( 1991 ) , the ellipticity @xmath211 and the @xmath80 fourier coefficients were kept constant within @xmath209 to their value at @xmath209 , and the surface - brightness profile was truncated at its value at @xmath208 . a new parametric bulge - disk decomposition was then performed on the truncated photometric data . the resulting new values of the effective radius and ellipticity of the bulge were in turn used to obtain a further estimate of @xmath208 and @xmath209 . the surface photometry was again modified according to these new values , and then another parametric fitting was done . the process was repeated up to convergence . a least - squares fit to @xmath216 in the bulge - dominated region beyond @xmath208 was performed using the 2-d brightness distribution resulting from the 3-d luminosity density given for a spherical body by : _ ( i ) _ a modified hubble law ( rood et al . 1972 ) : @xmath217^{-\frac{3}{2 } } \label{eq : hubble}\ ] ] where @xmath218 and @xmath219 are respectively the central luminosity density and the core radius ; _ ( ii ) _ a jaffe law ( jaffe 1983 ) : @xmath220 where @xmath221 and @xmath222 are the total luminosity and the half - light radius ; _ ( iii ) _ a hernquist law ( hernquist 1990 ) : @xmath223 where @xmath221 and @xmath224 are the total luminosity and a scale radius . the best fit was achieved for ngc 2179 with a hernquist profile and for ngc 2775 with a modified hubble profile . we used them to extrapolate the @xmath216 profiles of the two galaxies to @xmath225 . after subtracting the disk contribution from the total surface brightness , the 3-d luminosity density of the bulge was obtained starting lucy s iterations from a flattened hernquist model for ngc 2179 and from a flattened modified hubble model for ngc 2775 . the radial profiles of these flattened models are derived respectively from eqs . [ eq : hernquist ] and [ eq : hubble ] by replacing @xmath50 with @xmath226 where @xmath227 is the flattening and @xmath228 cylindrical coordinates . in this section we present the mass models of ngc 2179 and ngc 2775 . in fig . [ fig : n2179fot ] we show the @xmath78-band surface brightness ( @xmath229 ) , ellipticity ( @xmath211 ) and @xmath80 fourier coefficient of the isophote deviations from elliptical , as a function of radius along the major axis . the seeing cutoffs are @xmath230 and @xmath231 . the corresponding best - fit parameters obtained from the photometric decomposition are : @xmath232 @xmath71 , @xmath233 , @xmath234 for the de vaucouleurs bulge ; @xmath235 @xmath71 , @xmath236 , @xmath237 for the exponential disk . ( taking into account the photometric bulge - disk decomposition , the exponential disk yielding the best - fit model to the observed stellar kinematics has @xmath238 @xmath71 , @xmath239 kpc , and @xmath240 ; see dashed curve in fig . [ fig : n2179fot ] ) . we then subtract the disk contribution from the total surface brightness . the residual surface brightness is the contribution of the spheroidal component . the difference between the surface brightness of the spheroid and that obtained projecting the 3-d luminosity distribution of each of the four lucy iterations ( including the initial flatted hernquist model ) is shown in fig . [ fig : n2179dep ] ( right panel ) along ngc 2179 s major , minor and two intermediate axes . ( the r.m.s . residual of the last lucy iteration corresponds to 0.06509 ) . the 3-d luminosity density of the final bulge model along the same four axes is also presented in fig . [ fig : n2179dep ] ( left panel ) . we fold @xmath31 and @xmath56 around their respective centers of symmetry . in order to determine the latter , we fit a suitable odd function to both rcs independently : this yields the position of the kinematical center of the curve , @xmath241 , and the heliocentric velocity of the galaxy , @xmath3 . we find @xmath242 for both gas and stars , and @xmath243 and @xmath244 . we then fold @xmath32 and @xmath57 around the kinematical center of the respective component . the best - fit model to the observed major - axis stellar kinematics is shown in fig . [ fig : n2179mod_s ] . its parameters are as follows . the bulge is an oblate isotropic rotator ( @xmath245 ) with @xmath246 = 6.1 . the exponential disk has @xmath247 with scale - length @xmath248 kpc ) , and @xmath249 = 6.1 . the derived bulge and disk masses are @xmath250 and @xmath251 , adding up to a total ( bulge + disk ) luminous mass of @xmath252 . the dm halo has @xmath253 and @xmath254 kpc , which correspond to an asymptotic rotation velocity @xmath255 ; its mass at the outermost observed radius is @xmath256 . the ratio between the mass - to - light ratios of the stellar components in the models with and without the dm halo is 0.9 . the comparison between the observed rotation of the ionized gas and the true circular velocity , inferred from stellar kinematics , is given in the upper panel fig . [ fig : n2179mod_g ] . it shows that a dm halo is unambiguously required to explain the rotation at large radii ( @xmath257 ) . this result hinges on accuracy of the gas kinematics data beyond @xmath87 mostly derived from line measurements . in ngc 2179 , the gas rotation does provide the circular velocity at all radii . the contribution of the dm halo to ngc 2179 circular velocity as function of radius is plotted in lower panel of fig . [ fig : n2179mod_g ] . in fig . [ fig : n2775fot ] we show @xmath258 and @xmath259 along the major axis ( kent 1988 ) . the seeing fwhm for the kent ( 1988 ) data is @xmath260 . we derive the seeing cutoffs @xmath261 and @xmath231 . as no @xmath80 profile is available , we assume it is zero throughout . the photometric model is improved by taking into account an outer dust lane surrounding the spiral pattern of the galaxy . ( this appears as a thin dust ring at about @xmath102 from the center , and is visible on panels 78 and 87 of the cag , see sandage & bedke 1994 ) . as a fitting function for the disk component , we use an exponential profile weighted by an absorption ring . we assume the section of the dust ring to have a gaussian radial profile , defined by central intensity maximum absorption @xmath262 , a center @xmath263 and a scale - length @xmath264 . this leads to a disk light contribution @xmath265 the best - fit parameters resulting from the improved photometric decomposition are : @xmath266 @xmath50- , @xmath267 , @xmath268 for the de vaucouleurs bulge ; @xmath269 @xmath50- , @xmath270 , @xmath271 for the exponential disk ; and @xmath272 , @xmath273 and @xmath274 for the dust ring . the exponential disk yielding to the best - fit model to the observed stellar kinematics has @xmath269 @xmath50- , @xmath275 kpc , and @xmath276 ( see fig . [ fig : n2775fot ] ) . the surface brightness of the exponential disk in fig . [ fig : n2775fot ] is subtracted from the total surface brightness . the fit to the spheroidal component s deprojected surface brightness is obtained after four lucy iterations from an initial flatted modified hubble model ( the r.m.s . residual is 0.03209 , see fig . [ fig : n2775dep ] ) . the 3-d luminosity density of the final bulge model along the major , minor and two intermediate axes is also shown in fig . [ fig : n2775dep ] . we find @xmath31 and @xmath56 to have same center of symmetry at @xmath277 and @xmath278 . the velocity dispersion profiles are folded around the kinematical center . the best - fit model to the observed major - axis stellar kinematics is shown in fig . [ fig : n2775mod_s ] . its parameters are as follows . the bulge is an oblate isotropic rotator ( @xmath245 ) with @xmath279 = 5.2 . the exponential disk has @xmath280 with scale - length @xmath281 kpc ) , and @xmath282 = 7.0 . the derived bulge and disk masses are @xmath283 and @xmath284 , so the total ( bulge + disk ) luminous mass is @xmath285 . we kept the ratio between the bulge and disk @xmath183 fixed at the value 1.36 ( see kent 1988 ) . the dm halo has central density @xmath286 and core radius @xmath287 kpc , which correspond to an asymptotic rotation velocity @xmath288 ; its mass at the outermost observed radius is @xmath289 . ( the latter should be considered an upper limit because the inner kinematics can be explained with no dm halo ) . contrary to ngc 2179 in this case we kept the same mass - to - light ratios of the stellar components in the models with and without the dm halo . for @xmath290 the presence on the equatorial plane of the dust ring , which reduces the light contribution of the ( faster - rotating ) disk stars of a further @xmath291 , causes the observed stellar kinematics to be more affected by the ( slower - rotating ) bulge stars . in this picture as shown in fig . [ fig : n2775mod_s ] , our bulge model agrees with the observed drop in velocity and the rise in velocity dispersion . ngc 2775 is a dust - rich system , as can be inferred from its dust - to - hi mass ratio ( roberts et al . 1991 ) , which is @xmath292 times larger than the mean s0/sa values ( bregman , hogg & roberts 1992 ) . although the total luminous mass found by kent ( 1988 ) @xmath293 ( with @xmath6 mpc@xmath7 ) is in good agreement with ours , his mass decomposition differs from ours . he assumed only the bulge to have an analytical @xmath216 ( a de vaucouleurs law with @xmath232 , @xmath294 , @xmath295 ) , while the disk @xmath216 was taken to be the bulge - subtracted major - axis profile . kent s approach is opposite to ours . we assume the disk to have an analytical @xmath216 , and the bulge @xmath216 to be the residual surface brightness after subtracting the disk ( assuming no a priori analytical expression or fixed axis ratio ) . scaled to our assumed distance , the luminosity of kent s bulge is @xmath296 of ours , while our disk luminosity is @xmath297 of kent s . van der marel et al . ( 1991 ) studied the effects of a @xmath80 deviation on the kinematics of ngc 4261 . they found that changing the @xmath80 fourier component from zero to @xmath298 produces variations of @xmath299 in velocity dispersion and @xmath300 in rotation velocity . [ fig : n2775mod_a4 ] shows the stellar kinematics of ngc 2775 in the case of slightly disky ( @xmath301 ) and slightly boxy ( @xmath302 ) isophotes ( solid and dotted line , respectively ) . the disky model rotates faster and has a lower @xmath57 than the boxy model in the inner @xmath303 ( 1.9 kpc ) . the differences in @xmath83 and in @xmath57 between the two models are @xmath304 : this means that , in the observed range of values , a difference of @xmath305 in @xmath80 coefficients corresponds to a difference of @xmath306 in velocities and @xmath307 in velocity dispersions . however , these uncertainties are immaterial to our results on the mass structure of ngc 2775 . the comparison between @xmath94 and the true circular velocity inferred from the stellar kinematics is shown in the upper panel of fig . [ fig : n2775mod_g ] . it shows that a dm halo is not strictly required to explain the rotation at large radii ( @xmath308 ) . the contribution of the dm halo to ngc 2775 circular velocity as a function of radius is plotted in the lower panel of fig . [ fig : n2775mod_g ] . inside @xmath151 on the receding arm , @xmath309 and @xmath310 . this rules out the case that the gas kinematics is dominated by random motions , and leads us to speculate that we are looking at gas rotating on a non - equatorial plane . we suggest this is the signature of a past external acquisition ( possibly from the companion galaxy ngc 2777 ) of gas still not completely settled onto the disk plane . we have presented the ionized - gas and stellar kinematics , measured along the major axis , for a sample of six early - type spiral galaxies . ( due to the high values of @xmath32 in their inner regions , the gas rcs can not be used as circular - velocity curves . ) for ngc 2179 and ngc 2775 , we have modeled both the stellar and the gaseous kinematics to derive the mass contribution of the luminous and dark matter to the total potential , improving on the efforts by kent ( 1988 ) from gas kinematics alone . we have found that the innermost kinematics ( @xmath311 ) is very well and uniquely reproduced by taking into account the two luminous components . in the ( very luminous ) early - type spirals considered here , there is a large inner region in which ( essentially ) light traces the mass and the dm is a minor mass component . this is agreement with the ` weak ' maximum disk paradigm proposed by persic & salucci ( 1990 ) , but in disagreement with the claim by courteau & rix ( 1998 ) according to which in the most luminous spirals dm is a protagonist at essentially any radii . more in detail we have found that in ngc 2179 the combined stellar and gaseous rotation data ( measured out to about @xmath0 ) require the presence of a massive dark halo . in ngc 2775 , more luminous and massive than ngc 2179 , we can rule out a significant halo contribution out to @xmath312 . this result complies with the general trend of mass distribution known for later spirals ( persic et al . 1996 ) . salucci & persic ( 1997 ) , considering a large number of galaxies of mixed morphologies ( ellipticals , late spirals , dwarfs , and lsbs ) , have suggested that the halo structural parameters and the connection between the dark and the luminous matter show a strong continuity when passing from one hubble type to another . ellipticals , considered as luminous spheroids , and spirals , considered as luminous disks , are evidently very different systems , markedly discontinuous in terms of the distribution and global properties of the luminous matter . however , in the structural parameter space , ellipticals and spirals are contiguous , the main difference being that the former are more concentrated in both the dark and luminous components , probably due to the baryons dissipational infall being deeper in ellipticals than in spirals ( e.g. , bertola et al . 1993 ) . if so , it is hardly surprising that sa galaxies , being in some sense intermediate systems consisting of a luminous spheroid embedded in a luminous disk , fit in the regularity pattern of the dark - to - visible mass connection shared by ellipticals and spirals . in fig . [ fig : dm ] we plot our derived dark - to - visible mass ratios at the farthest measured radii for ngc 2179 and ngc 2775 ( filled circles ) onto the distribution , derived by salucci & persic ( 1997 ) for galaxies with the same visible mass . the agreement is good . even if the present result on early - type spirals is preliminary and without pretending to draw general conclusions from one particular case , it nevertheless seems to agree with the idea that , for galaxies of all morphological types , the dark - to - luminous mass ratio at any given radius depends only on the ( luminous ) mass of the galaxy . we are indebted to r.p . van der marel for providing his @xmath313 modeling software which became the basis of our modeling package . we also thank r. falomo for providing some photometric data reduction tools . we are most grateful to the vatican observatory research group for allocating the observing time . particular thanks go to r. boyle , s.j . for his help during the observing run at the vatt . the research of ap was partially supported by an _ acciaierie beltrame _ grant . jcvb acknowledges a grant from telescopio nazionale galileo and osservatorio astronomico di padova . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . bernacca , p.l . , perinotto , m. 1970 , contr . asiago , 239 , 1 bertola , f. , rubin , v.c . , zeilinger , w.w . 1989 , apj , 345 , l29 bertola , f. , bettoni , d. , rusconi , l. , sedmak , g. 1984 , aj , 89 , 356 bertola , f. , pizzella , a. , persic , m. , salucci , p. 1993 , apj , 416 , l45 bertola , f. , cinzano , p. , corsini , e.m . , rix , h .- w . , zeilinger , w.w . 1995 , apj , 448 , l13 bertola , f. , cinzano , p. , corsini , e.m . , pizzella , a. , persic , m. , salucci , p. 1996 , apj , 458 , l67 bertola , f. , cappellari , m. , funes s.j . , j.g . , corsini , e.m . , pizzella , a. , vega beltrn , j.c . 1998a , apj , submitted bertola , f. , corsini , e.m . , vega beltrn , j.c . , sarzi , m. , pizzella , a. , funes , j.g . 1998b , in preparation binney , j. , davies , r.l . , illingworth , g.d . 1990 , apj , 361 , 78 binney , j. , tremaine , s. 1987 , galactic dynamics . princeton university press , princeton bregman , j.n , hogg , d.e . , roberts , m.s . 1992 , apj , 387 , 484 cinzano , p. 1995 thesis , universit di padova cinzano , p. , van der marel , r.p . 1994 , mnras , 270 , 325 courteau , s. , rix , h .- w . 1998 , apj , in press [ astro - ph/9707290 ] de vaucouleurs , g. , de vaucouleurs , a. , corwin , h.g . , buta , r.j . , paturel , g. , fouqu , p. 1991 , third reference catalogue of bright galaxies . springer - verlag , new york ( rc3 ) fasano , g. 1990 , internal report of the padova astronomical observatory fillmore , j.a . , boroson , t.a . , dressler , a. 1986 , apj , 302 , 208 freeman , k.c . 1970 , apj , 160 , 811 gerhard , o.e . 1993 , mnras , 265 , 213 gerhard , o.e . , vietri , m. , kent , s.m . 1989 , apj , 345 , l33 hernquist , l. 1990 , apj , 356 , 359 hoffleit , d. , jaschek , c. 1982 , the bright star catalogue . yale university observatory , new haven ( bsc ) keel , w.c . 1996 , apjs , 106 , 27 kent , s.m . 1985 , apjs , 59 , 115 kent , s.m . 1988 , aj , 96 , 514 kormendy , j. , westpfahl , d.j . 1989 , apj , 338 , 752 jaffe , w. 1983 , mnras , 202 , 995 lauberts , a. , valentijn , e.a . 1989 , the surface photometry catalogue of the eso - uppsala galaxies . eso , garching bei mnchen osterbrock , d.e . , fulbright , j.p . , martel , a.r . , keane , m.j . , trager , s.c . 1996 , pasp , 108 , 277 peletier , r.f . , davies , r.l . , illingworth , g.d . , davis , l.e . , cawson , m. 1990 , aj , 100 , 1091 persic , m. , salucci , p. 1990 , mnras , 247 , 349 persic , m. , salucci , p. , stel , f. 1996 , mnras , 281 , 27 roberts , m.s . , hogg , d.e . , bregman , j.n . , forman , w.r . , jones , c. 1991 , apjs , 75 , 751 rood , h.j . , page , t.l . , kintner , e.c . , king , i.r . 1972 , apj , 175 , 627 rubin , v.c . , burstein , d. , ford , w.k . , jr . , thonnard , n. 1985 , apj , 289 , 81 salucci , p. , persic , m. 1997 , in : m. persic , p. salucci ( eds . ) asp conf . 117 , dark and visible matter in galaxies and cosmological implications , asp , san francisco , p. 1 sandage , a. , bedke , j. 1994 , the carnegie atlas of galaxies . carnegie institution , flintridge foundation , washington ( cag ) sandage , a. , tammann , g.a . 1981 , a revised shapley - ames catalog of bright galaxies . carnagie institution , washington ( rsa ) sargent , w.l.w . , schechter , p.l . , boksenberg , a. , shortridge , k. 1977 , apj , 212 , 326 satoh , c. 1980 , pasj , 32 , 41 tully , r.b . 1988 , nearby galaxies catalog . cambridge university press , cambridge van der marel , r.p . 1991 , mnras , 253 , 710 van der marel , r.p . , binney , j. , davies , r.l . 1990 , mnras , 245 , 582 van der marel , r.p . , franx , m. 1993 , apj , 407 , 525 wilson , r.e . 1953 , general catalogue of stellar radial velocities . carnagie institution , washington	we present the stellar and ionized - gas velocity curves and velocity - dispersion profiles along the major axis for six early - type spiral galaxies . two of these galaxies , namely ngc 2179 and ngc 2775 , are particularly suited for the study of dark matter halos . using their luminosity profiles and modeling their stellar and gaseous kinematics , we derive the mass contributions of the luminous and the dark matter to the total potential . in ngc 2179 we find that the data ( measured out to about the optical radius @xmath0 ) unambiguously require the presence of a massive dark halo . for the brighter and bigger object ngc 2775 , we can rule out a significant halo contribution at radii @xmath1 . although preliminary , these results agree with the familiar mass distribution trend known for late - type spirals of comparable mass . lllllllllllllc & & & & & & & & & & & & + & & & & & & & & & & & & & + & & & & & & & & & & & & & + ngc 2179&sa & .sas0 .. &13.22&170&51&2885@xmath210&2673&35.6&172.6 & 51 & 41 & 54 & 8.5 + ngc 2775&sa(r)&.sar2 .. &11.03&163&44&1350@xmath210&1180&15.7 & 76.1&128&100 & 77 & 6.5 + ngc 3281&sa & .sas2p*&12.70&138&69&3380@xmath210&3098&41.3&200.2 & 99 & 86 & 50 & 6.5 + ic 724&sa & .s .. 1 .. &13.4 & 60&55&5974@xmath210&5853&78.0&378.2 & 70 & 62 & 56 & 2.0 + ngc 4698&sa & .sas2 .. &11.46&170&70 & 992@xmath210 & 909&12.1 & 58.7&119 & 76&113 & 4.7 + ngc 4845&sa & .sas2./&12.10 & 75&72&1084@xmath210 & 980&13.1 & 63.5&150 & 87&100 & 4.0 + notes col.(2 ) : classification from rsa ( sandage & tamman 1981 ) . col.(3 ) : classification from rc3 ( de vaucouleurs et al . 1991 ) . col.(4 ) : total observed blue magnitude from rc3 except for ic 724 ( rsa ) . col.(5 ) : observed position angle . col.(6 ) : inclination from rubin et al . ( 1985 ) except for ngc 2179 ( tully 1988 ) . col.(7 ) : heliocentric velocity of the galaxy derived as center of symmetry of the gas rc . col.(8 ) : systemic velocity derived from @xmath3 corrected for the motion of the sun with respect of the local group by @xmath4 . col.(9 ) : distance obtained as @xmath5 with @xmath6 mpc@xmath7 . col.(11 ) : radius of the 25 @xmath8mag arcsec@xmath9 isophote from rc3 . col.(12 ) : radius of the farthest measured stellar velocity . col.(13 ) : radius of the farthest measured gas velocity . col.(14 ) : total integration time of the spectroscopic observation .
recent observations of the luminosity of type ia supernovae indicate [ 1 , 2 ] an accelerated expansion of the universe and lead to the search for a new type of matter which violate the strong energy condition @xmath1 . the matter consent responsible for such a condition to be satisfied at a certain stage of evolution of the universe is referred to as _ dark energy_. there are different candidates to play the role of the dark energy . the type of dark energy represented by a scalar field is often called _ quintessence_. the transition from a universe filled with matter to an exponentially expanding universe does not necessarily require the presence of the scalar field as the only alternative . in particular one can try another alternative by using an exotic type of fluid - the so - called chaplygin gas which obeys an equation of state like @xmath2 [ 3 ] where @xmath3 and @xmath4 are respectively the pressure and energy density and @xmath5 is a positive constant . subsequently the above equation was modified to the form @xmath6 with @xmath7 . this model gives the cosmological evolution from an initial dust like matter to an asymptotic cosmological constant with an epoch that can be seen as a mixture of a cosmological constant and a fluid obeying an equation of state @xmath8 . this generalized model has been studied previously [ 4 , 5 ] . + in the paper [ 4 ] , the flat friedmann model filled with chaplygin fluid has been analyzed in terms of the recently proposed `` _ statefinder _ '' parameters [ 6 ] . in fact trajectories in the @xmath9 plane corresponding to different cosmological models demonstrate qualitatively different behaviour . the statefinder diagnostic along with future snap observations may perhaps be used to discriminate between different dark energy models . the above statefinder diagnostic pair are constructed from the scale factor @xmath10 and its derivatives upto the third order as follows : @xmath11 where @xmath12 and @xmath13 are the hubble parameter and the deceleration parameter respectively . these parameters are dimensionless and allow us to characterize the properties of dark energy in a model independent manner . the parameter @xmath14 forms the next step in the hierarchy of geometrical cosmological parameters after @xmath12 and @xmath15 . + in section ii , all the discussions are valid in general for @xmath16 , but in section iii we have specifically considered the simple case of a spatially flat universe ( @xmath17 ) , which naturally follows from the simplest version of the inflationary scenario and is confirmed by recent cmb experiments [ 7 ] . in our present work we consider a more general modified chaplygin gas obeying an equation of state [ 8 ] @xmath18 this equation of state shows radiation era ( when @xmath19 ) at one extreme ( when the scale factor @xmath10 is vanishingly small ) while a @xmath0cdm model at the other extreme ( when the scale factor @xmath10 is infinitely large ) . at all stages it shows a mixture . also in between there is also one stage when the pressure vanishes and the matter content is equivalent to a pure dust . we have further described this particular cosmological model from the field theoretical point of view by introducing a scalar field @xmath20 and a self - interacting potential @xmath21 with the effective lagrangian @xmath22 in the paper of gorini et al [ 4 ] , it has been shown that for the simple flat friedmann model with chaplygin gas can equivalently described in terms of a homogeneous minimally coupled scalar field @xmath20 . in this case frw equations for chaplygin gas fits in barrow s scheme [ 9 ] . following barrow [ 10 ] , kamenshchik et al [ 3 , 11 ] have obtained homogeneous scalar field @xmath23 and a potential @xmath21 to describe chaplygin cosmology . in the section ii , we have obtained the corresponding expressions for a modified chaplygin gas as generalization of the previous work . + it has been possible to express @xmath21 in terms of the scale factor @xmath10 for arbitrary values of the constant @xmath24 and have explicitly shown how @xmath21 varies as the scale factor interpolates between radiation and @xmath0cdm stages for different values of the constant @xmath25 . in the next stage evolution of the model universe has been studied in the @xmath26 plane for the entire physically realistic history of the universe and compared to that for either pure chaplygin gas model or a generalized chaplygin gas model as given in [ 5 ] . the metric of a homogeneous and isotropic universe in frw model is @xmath27\ ] ] where @xmath10 is the scale factor and @xmath28 is the curvature scalar . + the einstein field equations are @xmath29 and @xmath30 where @xmath4 and @xmath3 are energy density and isotropic pressure respectively ( choosing @xmath31 ) . + the energy conservation equation is @xmath32 using equation ( 2 ) we have the solution of @xmath4 as @xmath33^{\frac{1}{1+\alpha}}\ ] ] where @xmath34 is an arbitrary integration constant . + due to complicated form of the equation of state ( 2 ) , the scale factor @xmath10 can not be solved from the einstein field equations ( 5 ) and ( 6 ) for arbitrary @xmath35 . however for @xmath17 , the solution of scale factor @xmath10 has the form @xmath36=\frac{\sqrt{3}}{2}(1+a)c^{x}t\ ] ] where @xmath37 and @xmath38 is the hypergeometric function . + now for small value of scale factor @xmath10 , we have @xmath39 which is very large and corresponds to the universe dominated by an equation of state @xmath40 . + also for large value of scale factor @xmath10 , @xmath41 which correspond to an otherwise empty universe with a cosmological constant @xmath42 . + for accelerating universe @xmath15 must be negative i.e. , @xmath43 i.e. , @xmath44 this expression shows that for small value of scale factor we have decelerating universe while for large values of scale factor we have accelerating universe and the transition occurs when the scale factor has the expression @xmath45^{\frac{1}{3(1+\alpha)(1+a)}}$ ] . + considering now the subleading terms in equation ( 8) at large values of @xmath46 , we can obtain the following expressions for the energy density and pressure : @xmath47 and @xmath48 equations ( 13 ) and ( 14 ) describe the mixture of a cosmological constant equal to @xmath42 with a matter whose equation of state is given by @xmath49 which for a pure chaplygin gas reduces to @xmath8 . + now we consider this energy density and pressure corresponding to a scalar field @xmath20 having a self - interacting potential @xmath21 . the lagrangian of the scalar field has been given in equation ( 3 ) . the analogous energy density @xmath50 and pressure @xmath51 for the scalar field are the following : @xmath52^{\frac{1}{1+\alpha}}\ ] ] and @xmath53^{\frac{1}{1+\alpha}}-b \left[\frac{b}{1+a}+\frac{c}{a^{3(1+a)(1+\alpha ) } } \right]^{-\frac{\alpha}{1+\alpha}}\ ] ] + hence for flat universe ( i.e. , @xmath17 ) we have @xmath54^{\frac{1}{1+\alpha}}-b \left[\frac{b}{1+a}+\frac{c}{a^{3(1+a)(1+\alpha ) } } \right]^{-\frac{\alpha}{1+\alpha}}\ ] ] and @xmath55^{\frac{1}{1+\alpha}}+\frac{1}{2}b \left[\frac{b}{1+a}+\frac{c}{a^{3(1+a)(1+\alpha ) } } \right]^{-\frac{\alpha}{1+\alpha}}\ ] ] from equation ( 18 ) we have the relation between scale factor @xmath10 and scalar field @xmath20 as @xmath56 from equation ( 19 ) we get @xmath57 @xmath58 since @xmath59 . so when @xmath60 i.e. , when @xmath61 the situations are completely different for @xmath62 and for @xmath63 . in the second case @xmath64 as @xmath60 , but in the first case @xmath65 as @xmath60 . this is why we get qualitatively two different pictures in two cases - as reflected in the figures 2 and 5 . in the other limit @xmath66 , @xmath67 for the above two cases . + for example , @xmath68 , @xmath69 as @xmath66 + and @xmath70 , @xmath71 as @xmath66 . + + fig.1 fig.2 + + fig.3 figs . 1 - 3 shows variation of @xmath20 and @xmath72 against @xmath46 and @xmath20 for @xmath19 and @xmath73 . + fig.4 fig.5 + + fig.6 figs . 4 - 6 shows variation of @xmath20 and @xmath72 against @xmath46 and @xmath20 for @xmath63 and @xmath74 . the graphical representation of @xmath20 against @xmath46 and @xmath21 against @xmath46 and @xmath20 respectively have been shown in figures 1 - 3 for @xmath19 and figures 4 - 6 for @xmath63 . from figures 1 , 4 we have seen that scalar field @xmath20 sharply falls when scale factor @xmath10 increases , both for @xmath19 and @xmath63 . in figure 2 , we see that potential function @xmath21 sharply decreases from extremely large value and thereafter it increases to a fixed value for @xmath19 . on the other hand for @xmath63 , the nature of @xmath21 against @xmath46 is as stated in the following . for small @xmath46 , @xmath21 varies insignificantly from zero value and then increases sharply to a constant value . in figure 3 , @xmath21 decreases slightly and then increases to infinitely large value for @xmath19 while for @xmath63 , in figure 6 , @xmath21 decreases to zero asymptotically starting from a extremely large value . so @xmath21 demonstrates completely different behaviour for two different values of @xmath63 and @xmath19 . further the figures show how @xmath21 varies with @xmath20 with different values chosen for @xmath24 . the statefinder parameters have been defined in equation ( 1 ) . trajectories in the @xmath26 plane corresponding to different cosmological models , for example @xmath0cdm model diagrams correspond to the fixed point @xmath75 . + for friedmann model with flat universe ( i.e. , @xmath17 ) , @xmath76 and @xmath77 so from equation ( 1 ) we get @xmath78 thus we get the ratio between @xmath3 and @xmath4 : @xmath79 for modified chaplygin gas , velocity of sound can be written as @xmath80 from equations ( 24 ) and ( 25 ) we get the relation between @xmath14 and @xmath81 : @xmath82 in the @xmath26 plane the above equation has only one asymptote parallel to @xmath81-axis namely , @xmath83 and the asymptote intersects the curve in only one point @xmath84 . + figure 7 shows the variation of @xmath81 with the variation of @xmath14 for @xmath19 and for @xmath85 . the portion of the curve on the positive side of @xmath81 which is physically admissible is only the values of @xmath14 greater than @xmath86 . the part of the curve between @xmath87 and @xmath88 with positive value of @xmath81 is not admissible ( we have not shown that part in the figure 7 ) because for the chaplygin gas under consideration we face a situation , where the magnitude of the constant @xmath5 becomes negative . it can be shown in the following way . + + fig.7 + fig . 7 shows the variation of @xmath81 against @xmath14 for different values of @xmath24 (= 0.6 , 1 ) and for @xmath19 . since we consider the range of @xmath25 as @xmath89 , we immediately get the inequality relation @xmath90 for all points lying in the part of the curve between @xmath87 and @xmath88 as stated above . in this range positive values of the parameter @xmath81 is explicitly given by @xmath91\ ] ] the above expression for @xmath81 is obtained by solving equation ( 27 ) . one must note that the other root containing negative sign before the square root symbol is to be left out since we consider only the positive magnitude of @xmath81 . + again from ( 25 ) and ( 2 ) , we get @xmath92 which in view of ( 29 ) leads to the relation @xmath93 one can show by a simple algebraic exercise that the right hand side of ( 30 ) is clearly negative for all points corresponding to @xmath94 and hence @xmath5 becomes negative , which is unacceptable in view of the chaplygin gas model we are considering . this upper limit for @xmath14 corresponds to @xmath95 for @xmath19 that is when we choose our universe to be a mixture of radiation and chaplygin gas . + thus the curve in the positive side of @xmath81 starts from radiation era and goes asymptotically to the dust model . but the portion in the negative side of @xmath81 represents the evolution from dust state ( @xmath96 ) to the @xmath0cdm model ( @xmath97 ) . thus the total curve represents the evolution of the universe starting from the radiation era to the @xmath0cdm model . in this work , we have presented a model for modified chaplygin gas . in this model we are able to describe the universe from the radiation era ( @xmath19 and @xmath4 is very large ) to @xmath0cdm model ( @xmath4 is small constant ) . so compare to chaplygin gas model , the present model describe universe to a large extent . also if we put @xmath98 with @xmath99 , then we can recover the results of the chaplygin gas model . in figure 7 , for @xmath26 diagram the portion of the curve for @xmath100 between @xmath87 to @xmath88 is not describable by the modified chaplygin gas under consideration . for example , if we choose @xmath101 then from the curve @xmath102 which corresponds to @xmath103 and hence we have from the equation of state , @xmath104 which is not valid for the specific chaplygin gas model considered here . at the large value of the scale factor we must have some stage where the pressure becomes negative and hence @xmath5 has to be chosen positive . it follows therefore that a portion of the curve as mentioned above should not remain valid . + + * acknowledgement : * + the authors are grateful to the referees for their valuable suggestions . they also thank iucaa for worm hospitality where the major part of the work was done . one of the authors ( u.d ) is thankful to csir ( govt . of india ) for awarding a senior research fellowship . + * references : * + + @xmath105 $ ] n. a. bachall , j. p. ostriker , s. perlmutter and p. j. steinhardt , _ science _ * 284 * 1481 ( 1999 ) . + @xmath106 $ ] s. j. perlmutter et al , _ astrophys . j. _ * 517 * 565 ( 1999 ) . + @xmath107 $ ] a. kamenshchik , u. moschella and v. pasquier , _ phys . b _ * 511 * 265 ( 2001 ) ; v. gorini , a. kamenshchik , u. moschella and v. pasquier , _ gr - qc_/0403062 . + @xmath108 $ ] v. gorini , a. kamenshchik and u. moschella , _ phys . d _ * 67 * 063509 ( 2003 ) ; u. alam , v. sahni , t. d. saini and a.a . starobinsky , _ mon . not . * 344 * , 1057 ( 2003 ) . + @xmath109 $ ] m. c. bento , o. bertolami and a. a. sen , _ phys . d _ 66 043507 ( 2002 ) . + @xmath110 $ ] v. sahni , t. d. saini , a. a. starobinsky and u. alam , _ jetp lett . _ * 77 * 201 ( 2003 ) . + @xmath111 $ ] a. benoit , p. ade , a. amblard et al , _ astron . astrophys . _ * 399 * l25 ( 2003 ) . + @xmath112 $ ] h. b. benaoum , _ hep - th_/0205140 . + @xmath113 $ ] j. d. barrow , _ nucl . phys . b _ * 310 * 743 ( 1988 ) . + @xmath114 $ ] j. d. barrow , _ phys.lett . b _ * 235 * 40 ( 1990 ) . + @xmath115 $ ] v. gorini , a. yu . kamenshchik , u. moschella , v. pasquier , _ phys.rev . d _ * 69 * 123512 ( 2004 )	in this paper we have considered a model of modified chaplygin gas and its role in accelerating phase of the universe . we have assumed that the equation of state of this modified model is valid from the radiation era to @xmath0cdm model . we have used recently developed statefinder parameters in characterizing different phase of the universe diagrammatically .
the super - kamiokande ( sk ) water - cherenkov experiment @xcite has recently confirmed @xcite , with high statistical significance , the anomalous flavor composition of the observed atmospheric neutrino flux , as compared with theoretical expectations @xcite . the flavor anomaly had been previously found in kamiokande @xcite and imb @xcite , and later in soudan2 @xcite , but not in the low - statistics experiments nusex @xcite and frjus @xcite . the recent sk data have also confirmed earlier kamiokande indications @xcite for a dependence of the flavor anomaly on the lepton zenith angle @xcite , which is correlated with the neutrino pathlength . the features of this dependence are consistent with the hypothesis of neutrino oscillations , which represents the most natural ( and perhaps exclusive ) explanation of the data @xcite . the oscillation hypothesis is also consistent with other recent atmospheric neutrino data , namely , the finalized sample of kamiokande upward - going muons @xcite , the latest muon and electron data from soudan2 @xcite , and the samples of stopping and through - going muons in macro @xcite . the sk atmospheric @xmath7 measurements , which are described in detail in several papers @xcite , conference proceedings @xcite , and theses @xcite , demand the greatest attention , not only for their intrinsic importance , but also for their interplay with other oscillation searches , including solar @xmath7 experiments @xcite and long baseline oscillation experiments at reactors @xcite and accelerators @xcite . in this work , we contribute to these topics by performing a comprehensive , quantitative , and accurate study of the sk atmospheric @xmath7 data in the hypothesis of three - flavor mixing among active neutrinos . we also include , within the same framework , the recent data from the chooz reactor experiment @xcite . more precisely , we consider the five sk angular distributions associated to sub - gev and multi - gev @xmath0-like and @xmath1-like events @xcite and to upward through - going muons @xcite , for a total of 30 accurately computed observables ( 5 + 5 + 5 + 5 + 10 zenith bins ) . first we study how such observables vary with the oscillation parameters , and then we fit them to the experimental data as measured in sk for an exposure of 33 kty ( 535 days ) @xcite . in the two - flavor mixing case , we confirm the results of the sk collaboration analysis , namely , that @xmath2 oscillations are preferred over @xmath3 , and that the no oscillation case is excluded with high confidence . in the three - flavor mixing case , we perform our analysis with and without the additional constraints imposed by the chooz reactor experiment . in both cases , the analysis favors a dominance of the @xmath2 channel . without the chooz constraints , the amplitudes of the subdominant @xmath3 and @xmath4 transitions can also be relatively large , indicating that , at present , current sk data do not exclude sizable @xmath5 mixing by themselves . after combining the chooz and sk data , the amplitudes of the subdominant transitions become smaller , but we show that they can still play a nonnegligible role both in atmospheric , solar , and long baseline laboratory experiments . the plan of our paper is as follows . in section ii we discuss the 30 sk observables used in the analysis , as well as the chooz measurement . in section iii we set the notation for our three - flavor oscillation framework . the two - flavor subcases are studied in sec . iv . in sec . v we perform the three - flavor analysis of sk data ( with and without the chooz constraints ) and discuss the results , especially those concerning @xmath5 mixing . in sec . vi we study the implications of our analysis for the neutrino oscillation phenomenology . we conclude our work in sec . vii , and devote appendixes a and b to the discussion of technical details related to our calculations and to the statistical analysis . some of our previous results on two- and three - flavor oscillations of solar @xcite , atmospheric @xcite , laboratory @xcite neutrino experiments and their combinations @xcite will be often referred to in this work . in particular , the @xmath8 analyses in @xcite summarize the pre - sk situation . however , we have tried to keep this paper as self - contained as possible . in this section we discuss expectations and data for five sk zenith distributions : sub - gev @xmath1-like and @xmath0-like , multi - gev @xmath1-like and @xmath0-like , and upward - going muons , with emphasis on some critical aspects of both theory and data that are often neglected . finally , we discuss the chooz reactor results . a basic ingredient of any theoretical calculation or montecarlo simulation of atmospheric @xmath7 event rates is the flux @xmath9 of atmospheric neutrinos and antineutrinos s and @xmath10 s . of course , we properly distinguish @xmath7 from @xmath10 in the input fluxes and in the calculations of cross sections and oscillation probabilities . ] as a function of the energy @xmath11 and of the zenith angle @xmath12 . the flux @xmath13 is unobservable in itself , and what is measured is the distribution of leptons @xmath14 after @xmath7 interactions , as a function of the lepton energy @xmath15 @xmath16 and lepton zenith angle @xmath17 @xmath18 . in fig . 1 we show the sum of theoretical @xmath7 and @xmath10 fluxes , as a function of the neutrino zenith angle @xmath12 , for selected values of the energy @xmath11 ( 1 , 10 , and 100 gev ) . the fluxes refer to the calculations of @xcite ( hkkm95 , solid lines ) and @xcite ( agls96 , dots ) without geomagnetic corrections ( so that the sign of @xmath19 is irrelevant in this figure ) . the upper and middle panels refer to muon and electron neutrinos , respectively , while the lower panel shows their ratio . several interesting things can be learned from this figure . for instance , the often quoted value @xmath20 for the muon - to - electron neutrino ratio clearly holds only for low - energy , horizontal neutrinos . this ratio increases rapidly as the neutrino energy increases and as its direction approaches the vertical . in fact , both @xmath21 and @xmath5 fluxes decrease towards the vertical ( see upper and middle panels ) , where the slanted depth in the atmosphere is reduced ; however , @xmath5 s are more effectively suppressed than @xmath21 s , due to their different parent decay chains . in addition , the greater the energy of the parents , the longer the decay lengths , the stronger the dependence of the @xmath22 flux ratio on the slanted depth and thus on the zenith angle @xmath12 . in other words , high - energy , vertical `` atmospheric @xmath7 beams '' are richer in @xmath21 s and , therefore , are best suited in searches for @xmath23 oscillations , where initial @xmath5 s represent the `` background '' ( see also @xcite ) . we anticipate that , in fact , multi - gev data are more effective than lower - energy ( sub - gev ) data in placing bounds on @xmath23 transitions . another consequence of the non - flat @xmath22 ratio is the appearance of distortions of the zenith distributions that , although related to _ vacuum _ neutrino oscillation , _ do not _ depend on neutrino pathlength - to - energy ratio @xmath24 @xcite . finally , notice in fig . 1 that the good agreement at low energies between the two reported calculations of @xmath22 is somewhat spoiled at high energies . this shows that the often - quoted uncertainty of @xmath25 for the @xmath22 ratio , which has been estimated in detail for _ low energy , integrated fluxes _ @xcite , does not necessarily apply to high - energy or differential fluxes . therefore , also the @xmath26 lepton event ratio might suffer of uncertainties larger than @xmath25 in some energy - angle bins . our empirical estimate of such errors in the statistical analysis is detailed in appendix b. more precise estimates of the relative @xmath21 and @xmath5 flux uncertainties are in progress @xcite . a new , ab initio , fully three - dimensional calculation of the atmospheric neutrino flux @xcite is also expected to shed light on these issues . concerning the overall uncertainty of the theoretical neutrino flux normalization , it is usually estimated to be 2030% . most of the uncertainty is associated to the primary flux of cosmic rays hitting the atmosphere . it is important , however , to allow also for energy - angle variations of such normalization ( e.g. , the sk analysis @xcite allows the spectral index to vary within @xmath27 ) . our approach to this problem is detailed in appendix b. here we want to emphasize that valuable information about the overall flux normalization can be obtained from more precise cosmic ray data from balloon experiments such as bess @xcite , caprice @xcite , and mass2 @xcite . the bess experiment has recently reported a relatively low flux of cosmic primaries @xcite , which , as we will see , might represent a serious problem for the oscillation interpretation of the sk data ( see also @xcite ) . on the other hand , the mass2 experiment can also measure the flux of primary protons and secondary muons _ at the same time _ @xcite , and might thus provide soon an important calibration of the theoretical flux calculations @xcite . therefore , it is reasonable to expect , in a few years , a reduction and a better understanding of the overall neutrino flux uncertainty , with obvious benefits for the interpretation of the atmospheric @xmath7 anomaly . in order to obtain measurable quantities ( e.g. , the lepton zenith angle distributions ) , one has to make a convolution of the neutrino fluxes with the differential cross sections and detection efficiencies ( see , e.g. , @xcite ) . we consider five zenith angle distributions of leptons : sub - gev muons and electrons , multi - gev muons and electrons , and upward through - going muons . concerning the calculation of the first four distributions , we have used the same technique used in @xcite for the old kamiokande multi - gev distributions . this approach makes use of the energy distributions of parent _ interacting _ neutrinos @xcite . concerning upward through - going muons , we improve the approach used in @xcite by including the zenith dependence of the sk muon energy threshold as given in @xcite ; we also use the same sk choice for the parton structure functions ( grv94 @xcite , available in @xcite ) and muon energy losses in the rock @xcite . for all distributions ( sg , mg , and up@xmath0 ) , we obtain a good agreement with the corresponding distributions simulated by the sk collaboration , as reported in appendix a ( to which we refer the reader for further details ) . in short , we can compute five distributions of sk lepton events as a function of the zenith angle @xmath17 , namely , sub - gev @xmath0-like and @xmath1-like events ( 5 + 5 bins ) , multi - gev @xmath0-like and @xmath1-like events ( 5 + 5 bins ) , and upward through - going muons ( 10 bins ) , for a total of 30 observables . few other analyses report explicit calculations of sub - gev and multi - gev zenith distributions ( see , e.g. , @xcite ) or upward - going muon distributions ( see , e.g. , @xcite ) in agreement with the sk simulations . other authors perform detailed calculations but use a reduced zenith information , as that embedded , e.g. , in the up - down lepton rate asymmetry @xcite ( see , e.g. , @xcite ) . the experimental data used in our analysis are reported in table i and table ii , together with the corresponding expectations as taken from the sk montecarlo simulations . the numerical values have been graphically reduced from the plots in @xcite and thus may be subject to slight inaccuracies . i reports the zenith angle distributions of sub - gev and multi - gev events collected in the sk fiducial mass ( 22.5 kton ) during 535 live days , for a total exposure of 33 kty . fully and partially contained multi - gev muons have been summed . only single - ring events are considered . the distributions are binned in five intervals of equal width in @xmath28 , from @xmath29 ( upward going leptons ) to @xmath30 ( downward going leptons ) . the total number of events in the full solid angle is also given . the quoted uncertainties for the data points are statistical . the statistical uncertainties associated to the sk montecarlo expectations originate from the finite simulated exposure ( 10 years live time @xcite ) . table ii reports the differential and total flux of upward through - going muons as a function of the zenith angle . data errors are statistical only . in this case , there is no statistical error for the sk theoretical estimates , which are derived from a direct calculation @xcite and not from a montecarlo simulation . it is useful to display the information in tables i and ii in graphical form . to this purpose , we take the central values of the theoretical expectations in tables i and ii as `` units of measure '' in each bin . in other words , all @xmath0 and @xmath1 event rates ( either observed , or calculated in the presence of oscillations ) are normalized to their standard ( i.e. , unoscillated ) expectations @xmath31 and @xmath32 . anomaly which are hidden in the @xmath26 ratio and emerge only when @xmath0 and @xmath33 rates are separated . ] the following notations distinguish the various lepton samples : @xmath34 .super - kamiokande sub - gev and multi - gev @xmath1-like and @xmath0-like atmospheric neutrino data , compared with the corresponding montecarlo simulations in different zenith angle ( @xmath17 ) bins @xcite . the numerical values are graphically reduced from the plots in @xcite . units : number of events . experimental exposure : 33 kty , corresponding to 22.5 kton fiducial mass @xmath35 535 live days . simulated exposure : 10 yr @xcite . errors are statistical only . the super - kamiokande montecarlo simulations refer to hkkm95 neutrino fluxes @xcite . [ cols="^,^,^,^,^,^,^ " , ] figure 2 compares theory and data for the total lepton rates . the mg@xmath0 data sample is further divided into fully contained ( fc ) and partially contained ( pc ) events ( notice that mg@xmath1 events are all fc ) . the total number of lepton events are displayed in the plane @xmath36 , so that the standard expectations ( mc ) correspond to the point @xmath37 for each data sample . the up and pc muons have no electron counterpart and are shown in the single variable @xmath38 ( upper strips ) . we attach a @xmath39 common uncertainty to the mc muon and electron rates ( large slanted error bar ) , and allow for a @xmath25 uncertainty in the @xmath26 ratio ( small slanted error bar ) . the three subfigures 2(a ) , 2(b ) , and 2(c ) correspond to three choices for the theoretical predictions ( mc ) : ( a ) standard mc expectations ; ( b ) mc rates multiplied by 1.2 ; and ( c ) mc rates multiplied by 0.8 . figure 2(a ) clearly shows that , with respect to the standard mc , sk observes a deficit of muons ( stronger for low - energy sg@xmath0 data and weaker for high - energy up@xmath0 data ) and , at the same time , an excess of electrons ( both in the sg and mg samples ) . in the hypothesis of neutrino oscillations , these indications would favor @xmath40 transitions , with a mass square difference low enough to give some energy dependence to the muon deficit . as far as _ total _ sk rates are concerned , this is a perfectly viable scenario . figure 2(b ) shows how the previous picture changes when one allows for an overall increase of the mc expectations ( say , @xmath41 ) . the relative excess of electrons disappears , while the muon deficit is enhanced . this situation is consistent with @xmath42 transitions , which leave the electron rate unaltered . the current sk data link rather strongly @xmath42 oscillations to an overall increase of the mc expectations : indeed , a good fit to the sg and mg data samples requires a mc `` renormalization '' by a factor @xmath43 @xcite . although this factor is acceptable at present , it might not be so in the future , should the mc predictions become more constrained . in particular , if the recent bess indications @xcite for a relatively low flux of cosmic primaries were confirmed , then one should rather _ decrease _ the mc expectations @xcite . figure 2(c ) shows the effect of a mc decrease by 20% . in this case one would observe no deficit of muons ( and even an excess of up@xmath0 events ) and a @xmath44 excess of electrons , which can not be obtained in any known oscillation scenario . the discussion of fig . 2 shows that : ( i ) theory ( no oscillation ) and data disagree , even allowing for a mc renormalization ; ( ii ) the oscillation interpretation depends sensitively on the size of the renormalization factor ; ( iii ) if this factor turns out to be @xmath45 , the oscillation hypothesis is jeopardized ; and ( iv ) it is thus of the utmost importance to calibrate and constrain the theoretical neutrino flux calculations @xcite through cosmic ray balloon experiments such as bess @xcite , and especially through simultaneous measurements of primary and secondary charged particles as in the forthcoming caprice and mass2 analyses @xcite . all this information would be lost if the popular `` @xmath26 double ratio '' were used . another piece of information that would be hidden by the double ratio is the fact that , in fig . 2 , the sg and mg data points appear to be very close to each other in the @xmath36 plane , while it was not so in kamiokande @xcite . we are , however , unable to trace the source of such a difference ( which is independent of the mc normalization ) between sk and kamiokande . since the total lepton rate information is subject to the above ambiguities , one hopes to learn more from _ differential rates _ and , in particular , from the zenith distributions of electrons and muons . these distributions are shown in fig . 3 where , again , the rates @xmath46 have been normalized in each bin to the central values of their expectations @xmath47 ( from tables i and ii ) . therefore , the no oscillation case corresponds to `` theory = 1 '' in this figure , with an overall normalization error that we set at @xmath48 . deviations of the data samples ( dots with error bars ) from the standard ( flat ) distribution are then immediately recognizable . the electron samples ( sg@xmath1 and mg@xmath1 ) do not show any significant deviation from a flat shape , with the possible exception of a slight excess of upward - going ( @xmath49 ) sg electrons . on the other hand , all the muon samples show a significant slope in the zenith distributions , especially for multi - gev data . , on average ) and therefore the flux of leptons is more diffuse in the solid angle . ] most of this work is devoted to understand how well two- and three - flavor oscillations of active neutrinos can explain these features of the sk angular distributions . we mention that additional sk measurements can potentially corroborate the neutrino oscillation hypothesis , namely , the stopping - to - passing ratio of upward - going muons @xcite , neutral - current enriched event samples @xcite , and azimuth ( east - west ) distributions of atmospheric @xmath7 events @xcite . such preliminary data will be considered in a future work . finally , the sk data themselves could be used for a self - calibration of the overall neutrino flux normalization . in particular , one should isolate a sample of high - energy , down - going leptons with directions close to the vertical , so that the corresponding parent neutrinos would be characterized by a pathlength , say , @xmath50 km and by an energy @xmath5120 gev . then , for a neutrino mass difference @xmath52 smaller than @xmath53 ev@xmath54 @xcite , the oscillating phase @xmath55 would also be small , and the selected sample could be effectively considered as unoscillated , thus providing a model - independent constraint on the absolute lepton rate and on the neutrino flux . the present sk statistics for strictly down - going , high - energy leptons , is not yet adequate to such a calibration . we will come back to this issue in the following . the chooz experiment @xcite searches for possible @xmath56 disappearance by means of a detector placed at @xmath57 km from two nuclear reactors with a total thermal power of 8.5 gw . with an average value of @xmath58 km / gev , it is able to explore the @xmath59 oscillation channel down to @xmath60 ev@xmath54 in the neutrino mass square difference , improving by about an order of magnitude previous reactor limits @xcite . the sensitivity to neutrino mixing is at the level of a few percent , being mainly limited by systematic uncertainties in the absolute reactor neutrino flux . the ratio of observed to expected neutrino events is @xmath61 , thus placing strong bounds on the electron flavor disappearance @xcite . the chooz limits have been recently retouched ( weakened ) @xcite as a result of the unified approach to confidence level limits proposed in @xcite . the impact of chooz for atmospheric oscillation searches and for their interplay with solar neutrino oscillations @xcite has been widely recognized ( see , e.g. , @xcite ) . earlier studies of the interplay between reactor , atmospheric , and solar neutrino experiments can be found in @xcite . given their importance , we have performed our own reanalysis of the chooz data in order to make a proper sk+chooz combination . we use the @xmath62 cross section as in our previous works @xcite , and convolute it with the reactor neutrino energy spectrum @xcite in order to obtain the positron rate . the expected rate is then compared with the data @xcite through a @xmath63 analysis . we have checked that , in the case of two - family oscillations , we obtain with good accuracy the exclusion limits shown in @xcite . our chooz reanalysis will be explicitly presented in section iv . the chooz bound counts as one additional constraint ( the observed positron rate ) ; therefore , the global sk+chooz analysis represents a fit to 30 + 1 observables . in this section we set the convention and notation used in the oscillation analysis . we consider three - flavor mixing among active neutrinos : @xmath64 with @xmath65 it is sometimes useful to parametrize the mixing matrix @xmath66 in terms of three mixing angles , @xmath67 , @xmath68 , and @xmath69 : @xmath70 where @xmath71 , @xmath72 , and we have neglected a possible cp violating phase that , in any case , would be unobservable in our framework . the mixing angles @xmath73 are also indicated as @xmath74 in the literature . while three - flavor oscillation probabilities are trivial to be computed in vacuum ( i.e. , in the `` atmospheric part '' of the neutrino trajectory ) , refined calculations are needed to account also for matter effects in the earth . as in our previous works @xcite , we solve numerically the neutrino evolution equations for any neutrino trajectory , taking into account the corresponding electron density profile in the earth . our computer programs are designed to compute the neutrino and antineutrino oscillation probabilities for _ any possible _ choice of values for the neutrino masses @xmath75 and mixing angles @xmath73 . a complete exploration of the three - flavor neutrino parameter space would be exceedingly complicated . therefore , data - driven approximations are often used to simplify the analysis @xcite . as in our previous works @xcite , we use the following hypothesis about neutrino square mass differences : @xmath76 i.e. , we assume that one of the square mass differences ( @xmath77 ) is much smaller than the other @xmath78 , which is the one probed by atmospheric neutrino experiments ( and accelerator or reactor experiments as well @xcite ) . the small square mass difference is then presumably associated to solar neutrino oscillations @xcite . notice that the above approximation involves squared mass differences and not the absolute masses ( which can not be probed in oscillation searches ) . in particular , eq . ( [ appr ] ) simply states that there is a `` lone neutrino '' @xmath79 , and a `` neutrino doublet '' @xmath80 , the doublet mass splitting being much smaller than the mass gap with the lone neutrino . however , eq . ( [ appr ] ) can be fulfilled with either @xmath81 or @xmath82 . these two cases are not entirely equivalent when matter effects are taken into account , as shown in @xcite . however , the difference is hardly recognizable in the current atmospheric @xmath7 phenomenology @xcite . for simplicity , in this paper we refer only to the case @xmath81 , i.e. , to a `` lone '' neutrino @xmath79 being the heaviest one . as far as @xmath83 ev@xmath54 and @xmath84 ev@xmath54 , atmospheric neutrino oscillations depend effectively only on @xmath85 . however , for larger values of @xmath77 the approximation ( [ appr ] ) begins to fail , and subleading , @xmath77-driven oscillations can affect the atmospheric @xmath7 phenomenology . we will briefly comment on subleading effects in sec . vi c. earlier discussions of such effects in solar and atmospheric neutrinos can be found in @xcite . under the approximation ( [ appr ] ) one can show that cp violating effects are unobservable , and that the angle @xmath67 can be rotated away in the analysis of atmospheric neutrinos ( see @xcite and references therein ) . in other words , atmospheric @xmath7 experiments do not probe the mixing @xmath86 associated with the quasi - degenerate doublet @xmath80 , but only the flavor composition of the lone state @xmath79 , @xmath87 the neutrino oscillation probabilities in vacuum assume then the simple form : @xmath88 where @xmath89\cdot l\,[{\rm km } ] } { e_\nu\ , [ { \rm gev}]}\right)\ , \label{s}\ ] ] neutrino and antineutrino probabilities being equal . however , in matter @xmath90 , and the probabilities must be calculated numerically for the earth density profile . for a constant density , they can still be calculated analytically @xcite . one can make contact with the familiar two - flavor oscillation scenarios in three limiting cases : @xmath91 with the corresponding , further identifications ( valid only in the @xmath92 cases ) : @xmath93 finally , we remind that for pure @xmath2 oscillations the physics is symmetric under the replacement @xmath94 , due to the absence of matter effects . such effects instead break the ( vacuum ) symmetry @xmath95 for pure @xmath3 oscillations , for which the cases @xmath96 and @xmath97 are distinguishable . and consider the cases @xmath98 or @xmath99 @xcite . ] in generic three - flavor cases , no specific symmetry exists in the presence of matter and the mixing angles @xmath68 and @xmath69 must be taken in their full range @xmath100 $ ] . a full account of the symmetry properties of the oscillation probability in @xmath92 and @xmath8 cases , both in vacuum and in matter , can be found in appendix c of @xcite . as previously said , under the hypothesis ( [ appr ] ) the parameter space of atmospheric @xmath7 s is spanned by @xmath101 . at any fixed value of @xmath85 , the unitarity condition @xmath102 can be embedded in a triangle graph @xcite , whose corners represent the flavor eigenstates , while a generic point inside the triangle represents the `` lone '' mass eigenstate @xmath79 . by identifying the heights projected from @xmath79 with the square matrix elements @xmath103 , @xmath104 , and @xmath105 , the unitarity condition ( [ unit ] ) is automatically satisfied for a unit height triangle @xcite . 4 shows the triangle graph as charted by the coordinates @xmath106 ( upper panel ) or @xmath107 ( lower panel ) . when @xmath79 ( the mass eigenstate ) coincides with one of the corners ( the flavor eigenstates ) , the no oscillation case is recovered . the sides correspond to pure @xmath92 oscillations . inner points in the triangle represent genuine @xmath8 oscillations . in this section we study first how the theoretical zenith distributions are distorted , in the presence of two flavor oscillations , with respect to the `` flat '' expectations of fig . 3 . this introductory study does not involve numerical fits to the data , and helps to understand which features of either @xmath2 or @xmath3 oscillations may be responsible for the observed sk zenith distributions . we then fit the most recent sk data ( 33 kty ) using a @xmath63 statistics , and discuss the results . figure @xmath108 shows , in the same format as in fig . 3 , our calculations of the five zenith distributions of atmospheric neutrino events . in this and in the following figures , the upper left box contains comments on the scenario , while the upper right box displays the selected values of @xmath109 . in fig . 5 we consider , in particular , pure @xmath2 oscillations @xmath110 with maximal mixing ( @xmath111 ) . of course , the electron distributions are not affected by @xmath2 transitions , while the muon event rates are suppressed , especially for zenith angles approaching the vertical ( @xmath29 , upward leptons ) , corresponding to longer average neutrino pathlengths . the prediction for @xmath112 ev@xmath54 ( dashed line ) is in reasonable agreement with all the muon data samples ( sg , mg , and up ) . for @xmath113 ev@xmath54 the expected rates of sg@xmath0 and up@xmath0 are significantly suppressed , and for @xmath114 ev@xmath54 one approaches the limit of energy - averaged @xmath92 oscillations , with a flat suppression of @xmath115 , which does not appear in agreement with the data . on the other hand , decreasing @xmath85 down to @xmath116 ev@xmath54 ( thick , solid line ) , one has almost `` unoscillated '' distributions for the high energy samples mg@xmath0 and up@xmath0 , since the phase @xmath117 is small . at lower energies ( i.e. , sg@xmath0 events ) , however , this phase can still be large enough to distort the zenith distribution . notice that in fig . 5 the theoretical electron distributions sg@xmath1 and mg@xmath1 are always below the data points . using the overall @xmath48 normalization freedom , one can imagine to `` rescale up '' all the five theoretical distributions ( by , say , 1520% ) to match the electron data . this upward shift would also alter the muon distributions at the same time , but one can easily realize that such distributions would still be in reasonable agreement with the data for @xmath112 and @xmath53 ev@xmath54 . therefore , at @xmath118 one expects an allowed range of @xmath85 around @xmath112@xmath53 ev@xmath54 , independently of the details of the statistical analysis . values of @xmath85 outside this range do not agree with the muon data . as a final comment to fig . 5 , we notice that the theoretical rate for downgoing multi - gev muons ( rightmost bin of the mg@xmath0 sample ) is practically identical to the unoscillated case for @xmath119 ev@xmath54 , as also observed at the end of sec . ii b. therefore , the sk data can potentially self - calibrate the _ absolute _ muon rate normalization _ independently of oscillations _ , provided that the total experimental error in the last mg@xmath0 bin is reduced to a few percent . in fig . 6 we take @xmath85 fixed ( at @xmath120 ev@xmath54 ) and vary the @xmath21-@xmath121 mixing ( @xmath122 ) . the suppression of the muon rates increases with increasing mixing ; however , there is no dramatic difference between @xmath118 and 0.75 ( solid and dashed lines , respectively ) . a value as low as @xmath123 is not in agreement with the sg and mg muon distributions , although it is still allowed by the up@xmath0 sample ( which does not place strong bounds on the mixing ) . if all the distributions were renormalized ( scaled up ) to match the sg@xmath1 and mg@xmath1 samples , small mixing values would be even more disfavored . therefore , we expect the mixing angle to be in the range @xmath1241 , independently on the details of the statistical analysis . we emphasize that , although in principle @xmath1-like events are not affected by @xmath2 oscillations , the sg@xmath1 and mg@xmath1 samples affect indirectly the estimate of the mass - mixing parameters , since they drive the fit to higher values of the neutrino fluxes . further experimental constraints on the overall neutrino flux normalization might have then a significant impact on the current estimates of @xmath85 and @xmath125 . 7 is analogous to fig . 5 , but for @xmath3 oscillations with maximal mixing ( @xmath126 and @xmath127 ) . matter effects are included . the expected sg and mg rates of electrons coming from below appear to be enhanced , due to @xmath21 s oscillating into @xmath5 s . the slope of the zenith distribution is stronger for mg@xmath1 than for sg@xmath1 , because : ( i ) the @xmath22 flux ratio increases with energy , as observed in section ii a ; and ( ii ) the `` angular smearing '' due to the different lepton and neutrino directions is more effective for sg events . on the other hand , the suppression of the muon rates is not as effective as for the @xmath2 case in fig . 5 . in fact , now there are some @xmath5 s oscillating back into @xmath21 s . moreover , matter effects tend to _ suppress _ large - amplitude oscillations ( when the mixing is maximal in vacuum , it can only be smaller in matter ) . in general , one has a too strong increase of electrons and a too weak suppression of muons , although this pattern may be in part improved by _ rescaling down _ all the theoretical curves . in this case , the distributions at @xmath112 and @xmath53 ev@xmath54 can get in marginal agreement with all the data , while @xmath128 ev@xmath54 is in any case excluded . notice , however , that @xmath113 ev@xmath54 is not allowed by chooz @xcite . figure 8 shows the effect of varying the @xmath3 mixing , for @xmath129 ev@xmath54a value safely below the chooz bounds . it can be seen that variations of the mixing do not help much in reaching an agreement with the data for this value of @xmath85 , which thus seems to be disfavored . figure 9 shows the size of matter effects for @xmath129 ev@xmath54 and maximal @xmath3 mixing . for values of @xmath85 around @xmath130 ev@xmath54 such as this , matter effects are more important for the multi - gev sample , while lower @xmath85 values would enhance the effect for the sub - gev sample and higher values for the up@xmath0 sample ( see , e.g. , @xcite and figs . 4 and 6 in @xcite ) . it can be seen that , in any case , the net matter effect is a decrease of the slope of the zenith distribution for both muons and electrons , i.e. , a suppression of the oscillation amplitude with respect to the `` pure vacuum '' case . the effect is not completely reduced to zero above horizon , due to the neutrino - lepton angular smearing . we summarize the content of figs . 79 by observing that the theoretical zenith distributions , in the presence of @xmath3 oscillations , are at most in marginal agreement with the sk data set . the agreement is somewhat improved by rescaling down the expectations . earth matter effects are sizable and can not be neglected in the analysis . in the previous two subsections we have presented quantitative calculations of the zenith distributions in selected two - flavor scanarios , and a qualitative comparison with the data . here we discuss the results of a quantitative @xmath92 fit to the sk data . figure 10 shows the results of our @xmath63 analysis of the sk data ( sg@xmath1 , sg@xmath0 , mg@xmath1 , mg@xmath0 , and up@xmath0 data combined ) . the panel ( a ) refers to @xmath2 oscillations ( @xmath131 ) in the plane ( @xmath85,@xmath132 ) . we find a minimum value @xmath133 for 28 degrees of freedom ( 30 data points minus 2 oscillation parameters ) , indicating a good fit to the data . is not very meaningful at present . we prefer to focus on confidence intervals . ] this is to be contrasted to the value @xmath134 for the no oscillation case , which is therefore excluded by the sk data with very high confidence . the quantitative limits on the mass - mixing parameters are consistent with the qualitative expectations discussed in section iv a. moreover , the allowed region is in agreement with the global analysis of pre - sk data shown in fig . 2 of @xcite . larger than the one reported by the kamiokande collaboration @xcite , presumably as a result of a different approach to the statistical analysis @xcite . see also the comments of @xcite about the kamiokande bounds in @xcite . ] our allowed range of @xmath85 in fig . 10(a ) is somewhat narrower than the range estimated by the sk collaboration @xcite . we have checked that the differences are largely due to the fact that only sg and mg were fitted in @xcite , while here we include also up@xmath0 data , which help to exclude the lowest values of @xmath85 ( see also @xcite ) . to a lesser extent , our different definition of @xmath63 ( see appendix b ) also plays a role . figures 10(b ) and 10(c ) refer to @xmath3 oscillations ( @xmath135 ) in the plane ( @xmath85,@xmath136 ) , for @xmath96 and @xmath97 , respectively ( the two cases being different , see sec . iii b ) . the minimum value of @xmath63 is now much higher ( 67.7 and 68.6 for panel ( b ) and ( c ) , respectively ) indicating that @xmath3 oscillations are disfavored by the sk data . this represents an important step forward with repect to pre - sk data , which did not distinguish significantly between @xmath2 and @xmath3 oscillations @xcite ( and , actually , showed a slight preference for the latter @xcite ) . in figs . 10(b , c ) , the c.l . limits around the minimum appear to be shifted to higher values of @xmath85 if compared to fig . 10(a ) , as a consequence of matter effects that suppress the effective mixing in the lowest range of @xmath85 ( see also @xcite ) . the @xmath3 allowed regions in fig . 10(b ) and 10(c ) have a relatively scarce interest . on the one hand , they represent a poor fit to the sk data themselves . on the other hand , they are excluded by the chooz reactor experiment . figure 10(d ) shows the chooz bounds , as derived by our own @xmath63 reanalysis , in good agreement with the limits shown in @xcite . such bounds are in contradiction with the @xmath3 allowed regions in fig . 10(b , c ) at 90% c.l . , ( although there might be a marginal agreement at 99% c.l . ) ; therefore , we do not make any attempt to combine sk+chooz data in a @xmath92 analysis . summarizing , our results for two - flavor oscillations are consistent with the @xmath92 analysis of the sk collaboration @xcite , namely : ( i ) the no oscillation hypothesis is rejected with high confidence ; and ( ii ) @xmath2 oscillations are largely preferred over @xmath3 . on our part , we add the following nontrivial statement : ( iii ) the present sk bounds on the @xmath2 mass - mixing parameters are in good agreement with those obtained from the global analysis of pre - sk data in @xcite ( including nusex , frjus , imb , and kamiokande sub - gev and multi - gev data ) . finally , we comment upon recent claims @xcite of inconsistencies within the sk data , under the hypothesis of @xmath2 oscillations . the argument goes as follows . if @xmath2 oscillations are assumed , the electron excess has to be adjusted by rescaling up the theoretical predictions . then the overall muon deficit becomes close to maximal ( @xmath115 ) for the sg and mg data ( see our fig . 2 , middle panel ) . such large suppression of the muon rate seems to suggest energy - averaged oscillations ( i.e. , large @xmath85 ) . on the other hand , the zenith distortions _ require _ energy - dependent oscillations ( i.e. , relatively small @xmath85 ) . some difference between the `` total rate '' and the `` shape '' information , although exacerbated by the semiquantitative calculations of @xcite , indeed exists but is not entirely new , as it has already been investigated by the sk collaboration @xcite . in fact , the slide no . 18 of @xcite shows the separate @xmath2 fits to the `` total rate '' and `` shape '' data , the former preferring values of @xmath85 tipically higher than the latter . however , in the same slide @xcite one can see a reassuring , large overlap between the two allowed ranges ( at @xmath137 c.l . ) for @xmath138 ev@xmath54 . this means that , within the present ( relatively large ) experimental and theoretical uncertainties , there is no real contradiction between different pieces of sk data , as also confirmed by our good @xmath2 global fit to the sk data . however , the above remarks , as well as our comments on the absolute event rates in sec . ii , should be kept in mind when new , more accurate experimental or theoretical information will become available . in this section we discuss in detail a three - flavor analysis of the sk and chooz data . in the first subsection we show representative examples of @xmath8 oscillation effects on the zenith distributions . in the second subsection we discuss some issues related to the @xmath24 variable . in the third and fourth subsections we report the results of detailed fits to the sk data , without and with the additional constraints from the chooz experiment , respectively . we remind that , in three flavors , the chooz mixing parameter @xmath139 can be identified with @xmath140 [ eq . ( [ p1 ] ) ] ; therefore , the chooz constraint @xmath141 , valid for @xmath142 ev@xmath54 @xcite , translates into either @xmath143 or @xmath144 . in a three - flavor language , two - flavor oscillations with maximal @xmath2 mixing are characterized by @xmath145 ( the center of the lower side in the triangle graph of fig . analogously , maximal @xmath3 mixing is characterized by @xmath146 ( the center of the right side of the triangle ) . a smooth @xmath8 interpolation between these @xmath92 cases can be performed by gradually increasing the value of @xmath103 from 0 to 1/2 , and decreasing the value of @xmath105 from 1/2 to 0 at the same time , the element @xmath104 being adjusted to preserve unitarity . this exercise is performed in fig . 11 for a relatively high value of @xmath85 ( @xmath147 ev@xmath54 ) . the thick and thin solid lines represent the zenith distributions for pure @xmath2 and pure @xmath3 oscillations with maximal mixing , respectively . the dashed and dotted lines represent intermediate @xmath8 cases , the first being `` close '' to @xmath2 ( with an additional @xmath148 admixture of @xmath5 ) , and the second being `` close '' to @xmath3 ( with an additional @xmath148 admixture of @xmath121 ) . although none of the four cases depicted in fig . 11 represents a good fit to all the sk data , and three of them are excluded by chooz , much can be learned from a qualitative understanding of the zenith distributions in this figure . in the presence of oscillations , the distributions @xmath149 in fig . 11 are roughly given by @xmath150 @xmath31 and @xmath32 being the unoscillated rates . for a relatively high value of @xmath85 as that in fig . 11 , the asymptotic regime of energy - averaged oscillations approximately applies , except for the rightmost bins of the mg and up@xmath0 distributions ( where the shorter pathlengths require higher @xmath85 s for reaching such regime ) . then , for pure @xmath2 oscillations with maximal mixing , one has @xmath151 , @xmath152 , and @xmath153 , so that @xmath154 and @xmath155 , as indicated by the thick , solid lines in fig . 11 . for pure @xmath3 oscillations with maximal mixing , one has @xmath156 , @xmath157 , and @xmath153 , so that @xmath158 and @xmath159 . for sub - gev data , the often - quoted value @xmath160 applies , so that @xmath161 and @xmath162 , as indicated by the thin , solid lines in the sg@xmath1 and sg@xmath0 panels of fig . 11 . for multi - gev data , however , the value @xmath163 is not constant , ranging from @xmath164 along the horizontal @xmath165 to @xmath166 along the vertical @xmath167 , see also fig . 1 ) . therefore , the ratio @xmath38 decreases slightly from @xmath168 ( horizontal ) to @xmath169 ( vertical ) , while the ratio @xmath170 increases significantly from @xmath171 ( horizontal ) to @xmath164 ( vertical ) . this is particularly evident as a `` convexity '' of the mg@xmath1 distribution in fig . 11 ( thin , solid line ) . such behavior is not confined to pure two - flavor oscillations , but is also present in genuine three - flavor cases , as indicated by the dashed and dotted lines in the mg@xmath1 panel of fig . this shows that , in the presence of @xmath172 mixing ( @xmath173 ) , the variations of the unoscillated ratio @xmath163 with @xmath28 induce distortions of the zenith distributions even in the regime of energy - averaged oscillations @xcite , contrary to naive expectations . notice that such distortions do not depend on @xmath24 , but on @xmath174 and @xmath11 separately through @xmath175 . another peculiar distortion , not dependent on @xmath24 , is related to a genuine three - flavor effect in matter @xcite . this effect is basically due to the splitting of the quasi - degenerate doublet @xmath80 in matter which , in the limit of large @xmath85 , leads to an effective square mass difference @xmath176 and thus to a subleading oscillation phase @xmath177 which does not depend on @xmath11 but only on @xmath174 @xcite . the main effect , relevant for atmospheric neutrinos , is to _ decrease _ the @xmath21 survival probability @xmath178 by an amount @xmath179 which , for a constant electron density @xmath180 , reads @xcite @xmath181 notice that the oscillation amplitude can be sizable only for large three - flavor mixing , while it disappears for two - flavor mixing ( i.e. , when one of the @xmath106 is zero ) , as indicated by the comparison of the dotted and thin solid lines in the mg@xmath0 and up@xmath0 panels of fig . the phase of @xmath179 ( the argument of @xmath182 ) can be rather large in the earth matter ( @xmath1836 mol/@xmath184 ) and is modulated by the neutrino zenith angle @xmath12 . this modulation is particularly evident in the genuine @xmath8 cases of the up@xmath0 panel in fig . 11 ( dotted and dashed lines ) , since upward through - going muons are highly correlated in direction with the parent neutrinos ( @xmath185 ) . the @xmath19 modulation of @xmath179 is increasingly smeared out in the lower energy mg and sg muon samples . we remind that in fig . 11 the dashed curves ( @xmath186 ) correspond to a @xmath8 `` perturbation '' of pure @xmath2 oscillations ( thick solid curves , @xmath187 ) , while the dotted curves ( @xmath188 ) correspond to a @xmath8 `` perturbation '' of pure @xmath3 oscillations ( thin solid curves , @xmath189 . by comparing the dashed and thick solid curves , it can be seen that @xmath8 perturbations of @xmath2 oscillations do not alter dramatically the zenith distributions . the opposite happens for the @xmath3 case ( dotted and thin solid curves ) . this pattern can be explained as an interference between vacuum and matter effects . more precisely , let us consider the case of large @xmath85 and @xmath190 . a @xmath8 perturbation of @xmath2 oscillations with maximal mixing can be parametrized by taking @xmath191 and @xmath192 ( the @xmath92 case being recovered for @xmath193 ) . the relevant oscillation probabilities are then @xmath194 , @xmath195 , and @xmath196 [ notice that @xmath179 is of @xmath197 , see eq . ( [ dp ] ) ] . equations ( [ app1],[app2 ] ) give the muon and electron rates , @xmath198 and @xmath199 , respectively . therefore , to the first order in @xmath200 , the electron rate does not vary , and the muon rate varies little since @xmath201 and @xmath179 partly cancel . conversely , a @xmath8 perturbation of @xmath3 oscillations with maximal mixing can be parametrized by taking @xmath202 and @xmath203 ( the @xmath92 case being recovered for @xmath193 ) . the relevant oscillation probabilities are then @xmath156 , @xmath204 , and @xmath205 . the muon and electron rates are now given by @xmath206 and @xmath207 , respectively . to the first order in @xmath208 , the electron rate decreases , and also the muon rate is suppressed , since the terms @xmath209 and @xmath179 have the same sign . therefore , adding some @xmath121 mixing ( @xmath210 ) to @xmath3 oscillations changes the predictions considerably ( generally in the direction of a better fit to the data ) . for this reason , we expect significant changes in the fit to sk data when moving continuously from pure @xmath3 oscillations to genuine @xmath8 cases . for simplicity , we have discussed the above three - flavor effects at large @xmath85 . values of @xmath85 lower than in fig . 11 are more interesting phenomenologically ( being less constrained by chooz ) but more difficult to understand qualitatively , since the oscillations are no longer energy - averaged , and @xmath92 , @xmath8 , vacuum , and matter effects are entangled . numerical calculations are required , and the results for @xmath211 and @xmath212 ev@xmath54 are shown in figs . 12 and 13 , respectively ( with mixing values chosen as in fig . 11 ) . the value @xmath129 ev@xmath54 is safely below the chooz bounds . in figs . 12 and 13 , the genuine @xmath8 cases ( dashed and dotted curves ) show an improved agreement with the data ( with respect to pure @xmath2 oscillations ) , since they give an excess of electrons without perturbing too much the muon distributions . although this advantage is not decisive at present , in view of the large uncertainties affecting the absolute normalization of the lepton rates , it might become crucial when such uncertainties will be reduced . if two - flavor @xmath2 oscillations @xmath213 were the true and exclusive explanation of the sk atmospheric data , then it would make sense to try to reconstruct the ( unobservable ) @xmath24 distribution of parent neutrinos from the lepton energies and directions . in fact , any distortion effect should be related to vacuum oscillations and thus to the ratio @xmath24 rather than to @xmath174 and @xmath11 separately . the theoretical and experimental @xmath24 distributions for sk can be found in @xcite . however , beyond the @xmath92 approximation , there are several oscillation effects that do not depend on @xmath24 . some of these effects , originating from @xmath5 mixing @xmath214 , have been described in the previous subsection . non-@xmath24 effects also arise in the presence of two comparable square mass differences ( i.e. , @xmath215 instead of @xmath216 ) , or if non - oscillatory phenomena contribute to partially explain the data . therefore , plots in the @xmath24 variable convey correct and unbiased information only under the hypothesis of pure @xmath2 oscillations . in all other cases , including our @xmath8 framework , such plots can not be used consistently . the following @xmath8 fits , as for the @xmath92 cases , make use of the zenith distributions and not of the reduced @xmath24 information . figure 14 shows our three - flavor fit to the sk data in the triangle graph , for values of @xmath85 decreasing from @xmath217 to @xmath218 ev@xmath54 . the curves at 90% and 99% c.l . correspond to an increase of @xmath63 by 6.25 and 11.36 above the global minimum . the results , shown separately for sub - gev , multi - gev , and upward - going muons in the first three columns of triangles , are then combined in the last column . the chooz data are excluded in this fit , in order to study what one can learn just from the sk data . for @xmath219 ev@xmath54 , the sub - gev data exclude all two - flavor oscillation subcases ( the triangle sides ) , but are consistent with genuine three - flavor oscillations at large @xmath103 ( @xmath220 ) . also multi - gev data are not in agreement with two - flavor oscillations , although the @xmath2 subcase ( lower side of the triangle ) can not be excluded at 99% c.l . multi - gev data are well fitted by genuine three - flavor oscillations , but in a range of @xmath103 different ( lower ) than for sub - gev data . the quality of the mg fits improves rapidly as one moves from the right side ( @xmath3 ) to the inner part of the triangle , as expected from the discussion of fig . 11 in sec . v a. the good @xmath8 fit to sg and mg data is mainly driven by the genuinely @xmath8 matter effects discussed in sec . v a [ see eq . ( [ dp ] ) and related comments ] . upward going muon data are much less constraining at 99% c.l.they allow any oscillation scenario . at 90% c.l . they disfavor : ( i ) pure or quasi - pure @xmath4 and @xmath40 oscillations ( left and right sides ) ; and ( ii ) large three - flavor mixing ( the central region of the triangle ) . large @xmath8 mixing is excluded because it suppresses and distorts too much the up@xmath0 distribution ( see fig . the regions allowed separately by sg , mg , and up data have no common intersection , and the combination of all the data is a null region ( last triangle ) . for @xmath221 ev@xmath54 , the increasing energy - angle dependence of the oscillation probability helps to fit the data better . therefore , the regions allowed in the triangle are larger for each of the three data sets , and a small allowed region appears at 99% c.l . in the combination . such region is enlarged for a lower value of @xmath85 ( @xmath222 ev@xmath54 ) and it appears also at 90% c.l . at @xmath223 ev@xmath54 . for the latter value of @xmath85 , all the data are consistent with @xmath2 oscillations ( the lower side ) , while the other two - flavor subcases ( the left and right sides ) are excluded . the mg data sample `` repels '' @xmath3 oscillations more strongly than sg data since , as observed in sec . ii a , higher - energy samples are characterized by a larger @xmath26 unoscillated ratio , and thus are more sensitive to the presence ( or absence ) of @xmath21-@xmath5 mixing . for @xmath224 ev@xmath54 , also the global combination of the data is consistent with @xmath2 oscillations , with large but not necessarily maximal mixing . large values of @xmath103 are also allowed , indicating that the sk data , by themselves , do not exclude large three - flavor mixing . we remark that the goodness of the fit improves rapidly when one moves from the right side inwards ( i.e , when one `` perturbs '' pure @xmath3 oscillations ) , while it changes more slowly when one moves from the lower side upwards ( i.e. , when one `` perturbs '' pure @xmath2 oscillations ) , as expected from the discussion in sec . v a. figure 15 is analogous to fig . 14 , but for lower values of @xmath85 , ranging from @xmath225 to @xmath226 ev@xmath54 . as @xmath85 decreases , the sg data fit is not affected very much , since values as low as few@xmath227 ev@xmath54 still provide a good fit to this sample . however , the oscillation phase starts decreasing more rapidly for higher - energy samples ( mg and up ) , leading to an insufficient suppression of the muon rates and to a gradual reduction of the allowed regions . notice , in particular , how up@xmath0 data constrain @xmath8 mixing for @xmath2283 ev@xmath54 . in any case , the preferred regions are more and more reduced and closer to the lower side of the triangle , corresponding to smaller allowed values for @xmath103 . there is no joint allowed region at 99% c.l . for @xmath85 below @xmath229 ev@xmath54 . we summarize the three - flavor fit to sk data as follows : ( i ) the sk data exclude both @xmath230 and @xmath231 , being consistent with large @xmath2 mixing ( not necessarily maximal ) ; ( ii ) values of @xmath103 as large as 0.5 can not be excluded only on the basis of sk data . indeed , @xmath8 oscillations with large @xmath5 mixing can improve the fit and , in particular , it can explain ( part of ) the electron excess in the sg and mg samples . quantitative bounds on the mixing matrix elements @xmath232 can be derived from figs . 14 and 15 . in the next subsection we study the impact of chooz on such indications . we combine sk and chooz data ( 30 + 1 observables ) through a joint @xmath63 analysis . the results are reported in fig . figure 16 shows the 90% and 99% c.l . in the triangle plot for selected values of @xmath85 , ranging from @xmath233 ev@xmath54 ( upper triangles ) to @xmath234 ev@xmath54 ( lower triangles ) the left column of triangles reports the fit to sk data only ( as derived from figs . 14 and 15 ) . the middle column reports the fit to chooz data , which exclude a large horizontal stripe . in fact , the nonobservation of @xmath5 disappearance implies that @xmath79 is either very close to the upper corner @xmath5 ( so as to suppress oscillations ) or very close to the lower side ( @xmath2 oscillations being unobservables in chooz ) . clearly , the addition of the chooz bounds to the sk fit ( right column of triangles ) cuts significantly the upper part of the solutions , so that only relatively low values of @xmath103 are allowed . however , the chooz bounds are rapidly weakened as @xmath85 is decreased , and for @xmath235 ev@xmath54 the parameter @xmath103 can be as large as @xmath236 at 90% c.l.and @xmath237 at 99% c.l . , corresponding to a significant @xmath5 appearance probability [ see eq . ( [ p ] ) ] . therefore , the chooz data constrain but _ do not _ exclude the role of @xmath5 mixing and electron appearance in the interpretation of the sk data ( see also @xcite ) . in particular , part of the electron excess in the sg and mg samples could be explained by nonzero values of @xmath103 rather than by uncertainties in the overall neutrino flux normalization . nonzero values of @xmath103 also contribute to distort the zenith distributions @xcite , as discussed in sec . v a. so far we have seen the impact of chooz on the mixing parameters @xmath238 . the impact on the square mass difference @xmath85 is summarized in fig . 17 , which shows the @xmath63 as a function of @xmath85 , for unconstrained values of the mixing angles and for both fits to sk ( dashed lines ) and sk+chooz ( solid lines ) . the minimum value of @xmath63 is @xmath239 for sk and @xmath240 for sk+chooz , indicating a good fit to the data ( 30 and 31 observables , respectively ) . the chooz data help to constrain @xmath85 on the higher range , but its role decreases rapidly for @xmath241 ev@xmath54 . also shown in fig . 17 are the 90% and 99% c.l . intervals for @xmath85 , which allow values as low as @xmath242 ev@xmath54 . the possibility of exploring such low values of @xmath85 should be seriously considered in long baseline experiments . an interesting result of fig . 17 is the stability of the @xmath85 range indicated by sk it does not change dramatically by adding the chooz constraint . therefore , the inclusion of the chooz data in the global analysis affects more the mixing than the mass parameter . a comparison of sk and pre - sk bounds is illuminating . figure 17 should be compared with fig . 10 of @xcite , where we combined the data from nusex , frjus , imb , and kamiokande ( sub - gev and multi - gev ) in a three - flavor analysis . the comparison shows that the sk and sk+chooz bounds on @xmath85 are perfectly consistent with the pre - sk bounds . the sk+chooz data appear to improve significantly the old upper bound on @xmath85 , but give a lower bound very similar to the pre - sk data . notice that we have long since claimed that the popular value @xmath243 ev@xmath54 overestimated the best fit for pre - sk data , and that values as low as @xmath242 ev@xmath54 were compatible with the atmospheric @xmath7 data @xcite . we plan to perform a joint analysis of all the data ( sk+chooz+pre - sk ) in a future work . in this section we examine some implications of our three - flavor analysis for the phenomenology of atmospheric , long - baseline , and solar neutrino experiments , as well as for model building . the sk atmospheric data are consistent with @xmath8 oscillations with dominant @xmath2 transitions ( @xmath244 at 90% c.l . ) and subdominant @xmath245 mixing @xmath246 . the mass square difference @xmath85 is favored in the range @xmath247@xmath248 ev@xmath54 . can one improve significantly such indications only with sk or other atmospheric data ? large @xmath2 mixing should generate a @xmath121 flux comparable to the @xmath21 flux at the detector site . however , the `` contamination '' of @xmath0-like and @xmath1-like events from @xmath249 production and subsequent leptonic decay is estimated to be very small in sk @xcite . therefore , there seems to be little hope to test the @xmath2 channel through @xmath249 appearance in sk . nevertheless , the possibility of enhancing ( through appropriate cuts ) the @xmath250 and @xmath251 `` pollution '' in selected @xmath0 and @xmath1 event samples may deserve further attention in other atmospheric @xmath7 detectors such as soudan2 . the tests of @xmath5 mixing ( i.e. , of the matrix element @xmath103 ) can certainly be improved with higher statistics sk data , in particular with more multi - gev upgoing electron and muon events . such data samples are characterized by a relatively high @xmath22 flux ratio ( see sec . ii a ) , and thus are more sensitive to an increase of @xmath1-like events due to @xmath3 transitions . multi - gev data are already more powerful than sg ( and up@xmath0 ) data in constraining @xmath103 ( see fig . while the elements @xmath106 determine the _ amplitude _ of oscillations , which can be already derived from total event rates , the parameter @xmath85 governs the _ phase _ of oscillations , and thus it can be derived only through event spectra . hypothetical spectra of neutrino events as a function of @xmath11 and @xmath174 would be the most sensitive probes of @xmath85 . unfortunately , a complete kinematical closure of @xmath7-induced events can not be achieved in sk , so neither @xmath11 nor @xmath174 can be precisely reconstructed , especially for low - energy events . this intrinsic feature will eventually limit the maximum accuracy of @xmath85 fits attainable with sk data only . in this respect , the possibility of improving the @xmath11 and @xmath174 reconstruction in experiments as soudan2 @xcite ( through observation of the struck nucleon ) , or in high - density detectors as proposed in @xcite , appears extremely interesting and promising . long baseline ( lbl ) accelerator experiments , such as k2k @xcite , minos @xcite , and various cern - gran sasso proposals @xcite , are expected to confirm the atmospheric neutrino signal with a controlled beam . since both two - flavor @xcite and three - flavor analyses like ours show that @xmath85 can be as low as @xmath252 ( 99% c.l . ) , the design of low - energy beams should be pursued seriously . if the atmospheric @xmath85 range can be covered completely , then it suffices to have either a @xmath21 disappearance or a @xmath121 appearance signal to confirm the sk anomaly . however , we think that long baseline experiments should be designed to _ measure _ oscillation parameters , rather than merely _ to confirm _ an oscillation effect already found by sk . measuring the oscillation parameters is a task that demands careful considerations , especially if @xmath8 oscillations are to be tested ( for @xmath92 oscillations , see the lucid discussion in @xcite ) . the determination of @xmath85 requires energy spectra analyses , and thus high - statistics event samples . if @xmath85 happens to be in the low range of the experimental sensitivity , the @xmath249 appearance sample might consist of just a handful of events . the @xmath1 appearance event sample might also be small if @xmath253 . therefore , a safe reconstruction of @xmath85 should be based mainly on @xmath0 event spectra from the @xmath254 disappearance channel , where most of the signal is expected in any case . this implies a good monitoring of the initial @xmath21 beam with a near detector . the determination of the matrix elements @xmath106 requires that several oscillation channels are probed at the same time redundancy is never enough to constrain neutrino mixing @xcite . for instance , the @xmath254 disappearance channel is sensitive only to @xmath255 but tells nothing on @xmath256 or @xmath257 , while the @xmath258 appearance channel is sensitive to the product @xmath259 but it can not separate the two factors @xmath260 and @xmath261 nor measure @xmath103 [ see eqs . ( [ p1],[p ] ) ] . these aspects of @xmath8 mixing tests @xcite in long - baseline experiments are better appreciated in fig . figure 18 shows , in the first column of triangles , the region allowed by sk+chooz for selected values of @xmath85 . the second column shows , superimposed , the prospective regions that can be probed at @xmath262 c.l . by k2k @xcite , both in the @xmath254 channel ( slanted bands ) and in the @xmath40 channel ( hyperbola ) . in the @xmath263 disappearance channel or of the form @xmath264 in the @xmath265 appearance channel @xcite . ] it appears that k2k might not reach , in the @xmath40 channel , sufficient sensitivity to probe the values of @xmath103 allowed by sk+chooz . the disappearance channel @xmath254 can cover the whole sk+chooz region , but only for @xmath142 ev@xmath54 . therefore , k2k is basically expected to give information on @xmath255 for @xmath266 ev@xmath54 , given the sensitivities prospected in @xcite . with respect to k2k , the minos experiment is being designed to probe lower values of @xmath85 and to explore also the @xmath42 appearance channel ( the region below the hyperbola touching the @xmath42 side in fig . possible signals in the three channels @xmath267 will constrain the quantities @xmath268 , @xmath255 , and @xmath269 , respectively , so that the elements @xmath232 can be pinpointed for @xmath270 ev@xmath54 if the uncertainties are kept small . for lower @xmath85 s , minos rapidly looses sensitivity in at least one of the oscillation channels , and it might be difficult to constrain the neutrino mixing parameters . notice that , for @xmath271 and the @xmath272 ( allowed by sk+chooz ) , the @xmath40 appearance probability is @xmath273 , where @xmath274 is the oscillation factor in eq . ( [ s ] ) , averaged over the @xmath7 beam energy spectrum . depending on @xmath275 and on the specific mixing parameters , values of @xmath276 as large as 15% appear possible in properly designed lbl experiments . a final remark is in order . the sensitivity regions in fig . 18 have been derived from the prospective estimates reported in the experiment proposals @xcite , which are in continuous evolution ( even more so for the cern to gran sasso proposals @xcite , not shown ) . therefore , the above considerations on k2k and minos are to be considered as preliminary and qualitative . nevertheless , it remains true that lbl experiments might face some difficulties in constraining the @xmath8 mixing parameters , especially if @xmath85 is low or if the three oscillation channels @xmath267 can not all be probed . re - directing the goal of lbl experiments from `` confirming the super - kamiokande signal '' to `` measuring the parameters @xmath277 '' would be beneficial to the current debate on various lbl proposals . in the limit @xmath216 [ eq . ( [ appr ] ) ] , experiments with terrestrial ( atmospheric , accelerator and reactor ) neutrino beams probe the parameters @xmath278 . in the same approximation , solar neutrino experiments probe the parameters @xmath279 , i.e. , the small mass square difference and the mass composition of @xmath5 @xcite . therefore , both terrestrial and solar @xmath7 experiments probe the element @xmath103 , which parametrizes the mixing of @xmath5 with the `` lone '' neutrino mass eigenstate @xmath79 . the effect of such mixing on solar neutrinos is to give an energy - independent contribution to the disappearance of @xmath5 s from the sun . since this contribution is basically proportional to the square of @xmath103 @xcite , sizable mixing is required to have large effects . however , as @xmath103 is constrained by the sk+chooz fit , such effects are relatively small to be detected in the current solar neutrino experiments . the smallness of @xmath103 also reduces the `` coupling '' of the terrestrial and solar parameter spaces @xcite . nevertheless , it is interesting to notice that a preference for small values of @xmath103 emerges naturally from solar neutrino data only , in both matter - enhanced @xcite and vacuum @xcite three - flavor oscillation fits . an updated analysis using the latest sk solar neutrino data would be desirable to confirm such indication . in fact , a more pronounced preference of solar @xmath7 s for relatively small values of @xmath103 would be a nontrivial , although indirect , hint that both solar and terrestrial @xmath7 data are consistent within the same @xmath8 oscillation scenario . the relation between solar and atmospheric neutrinos is not necessarily confined to a preference for relatively small values of the mixing matrix element @xmath103 . for instance , if the assumption in eq . ( [ appr ] ) is violated , atmospheric neutrinos can become sensitive to the subleading oscillations driven by @xmath77 , i.e. , by the `` solar neutrino '' mass difference . figure 19 shows a representative case beyond the @xmath216 approximation . the values chosen for the oscillation parameters @xmath280 are allowed at 99% c.l . by the present sk+chooz bounds and by the pre - sk , three - flavor solar @xmath7 analysis in @xcite . the effect of the subdominant mass scale @xmath281 can be evaluated by comparing the solid curves ( @xmath282 ) and the dashed curves ( @xmath283 ) . of course , the effect is more significant when @xmath284 is larger , i.e. , for upgoing sg events , while it rapidly decreases at higher energies ( mg and up@xmath0 samples ) and shorter pathlengths ( @xmath285 ) . for sg@xmath0 events , however , the effect approaches the size of the error bars and thus it might be probed by sk ! however , the strategies for disentangling the oscillations driven by @xmath85 and @xmath77 are nontrivial and will be investigated in a future work . theoretical or phenomenological models of neutrino mass and mixing try to predict , or `` explain , '' the set of parameters @xmath75 and @xmath286 . our analysis constrains the subset of parameters @xmath287 , provided that @xmath77 is sufficiently small ( @xmath288 ) . many models that try to explain solar+atmospheric @xmath7 data fall within this category , and are thus strongly constrained by the sk+chooz bounds worked out in this paper . for instance , the so - called bimaximal mixing model @xcite , characterized by @xmath289 for atmospheric neutrinos [ and by @xmath290 for solar neutrinos ] is allowed for @xmath291 ev@xmath54 ( see fig . conversely , the trimaximal mixing model @xcite , characterized by @xmath292 for atmospheric neutrinos [ and by @xmath293 for solar neutrinos ] appears to be strongly disfavored by our combined sk+chooz analysis ( it would correspond to the center of each triangle in figs . 14 and 15 ) . of course , many other models ( see , e.g. , the classification in @xcite ) can be tested through our bounds on the oscillation parameters , provided that the dominance of the mass scale @xmath85 is assumed for atmospheric neutrinos . other models try to explain the atmospheric anomaly and the lsnd evidence @xcite for @xmath40 oscillations , with an allowance for an energy - averaged suppression of the solar neutrino flux . arguments disfavoring such scenarios are discussed in @xcite ( see , in particular , table vi of @xcite and related comments ) . one such model has been recently proposed in @xcite , where @xmath294 ev@xmath54 is assumed to drive the oscillations in the lsnd experiment @xcite range , as well as energy - averaged oscillations of atmospheric @xmath7 s , while @xmath295@xmath130 ev@xmath54 is assumed to drive energy - dependent oscillations of atmospheric @xmath7 s . both @xmath77 and @xmath85 can then contribute to the solar neutrino deficit through energy - averaged oscillations . since the bounds worked out in sec . v d assume @xmath296 ev@xmath54 , and thus do not apply to such model , we have performed a numerical analysis of sk data _ ad hoc _ , using the same mass - mixing parameters as in @xcite . we find that the resulting zenith angle distributions of muons ( not shown ) are only mildly distorted , and that the model is disfavored by the sk atmospheric data at @xmath297 c.l . , with or without matter effects . we mention that , for the choice of parameters in @xcite , matter effects influence significantly the zenith distributions , making them flatter than in vacuum . the semi - quantitative calculations in @xcite showed a more optimistic agreement to the sk data , in part because matter effects were ignored . in addition , the large value for @xmath257 chosen in @xcite does not appear in agreement with the global analysis of laboratory neutrino oscillation searches ( including the lsnd data ) performed in @xcite . we have performed a three - flavor analysis of the sk atmospheric neutrino data , in a framework characterized by the mass - mixing parameters @xmath277 , in the hypothesis of one mass scale dominance . the variations of the zenith distributions of @xmath7 events in the presence of flavor oscillations have been investigated in detail . fits to the sk data , with and without the additional chooz data , strongly constrain the parameter space . detailed bounds have been shown in triangle graphs , embedding the unitarity condition @xmath298 . the allowed regions include the subcase @xmath187 , corresponding to pure @xmath2 oscillations . however , values of @xmath299 are also allowed . in particular , for @xmath85 close to ( or slightly below ) @xmath130 ev@xmath54 , @xmath103 can be as large as @xmath236 ( at 90% c.l . ) . scenarios with @xmath300 correspond to genuine three - flavor oscillations and are characterized by a rich phenomenology , not only for atmospheric @xmath7 s , but also for solar and laboratory neutrino oscillation searches . in particular , challenging opportunities are disclosed for @xmath5 appearance searches in long baseline experiments . our analysis also places strong constraints on models of neutrino mass and mixing . in addition , we have examined many facets of the sk data and of their interpretation , that will deserve further attention when the experimental and theoretical uncertainties will be reduced . is grateful to h. minakata and to o. yasuda for fruitful discussions and for kind hospitality at tokyo metropolitan university . we thank d. montanino for helpful discussions . the work of a.m.and g.s . is supported by the ministero delluniversit e della ricerca scientifica ( dottorato di ricerca ) . the calculation of the zenith angle distributions of sg and mg lepton events involves the numerical evaluation of multiple integrals of the form @xmath301 where @xmath9 is the unoscillated neutrino spectrum , @xmath302 is the differential cross section for lepton production , @xmath303 is the detector efficiency for lepton reconstruction , and @xmath304 is the oscillation probability @xcite . the efficiency function @xmath303 is not always reported in the experimental papers . in particular , it has not been explicitly given by the sk collaboration so far . we faced a similar problem in the analysis of the kamiokande multi - gev data performed in @xcite . our solution @xcite was to use the energy distribution of the parent neutrinos @xmath7 that produce a _ detected _ lepton @xmath305 , namely , @xmath306 where @xmath15 is the lepton energy . this distribution , which gives information on the factor @xmath307 in eq . ( [ multi ] ) , has been published in @xcite for the kamiokande experiment . concerning sk , we have used the analogous information from @xcite . using the energy distribution of parent neutrinos , it is possible to reconstruct the zenith distribution of the final leptons , provided that the smearing induced by neutrino - lepton scattering angle is taken into account @xcite . while for the old kamiokande multi - gev data we approximated this effect with an energy - independent smearing angle of @xmath308 , for sk we properly take into account the distribution of the lepton scattering angle and its dependence on the energy , which is especially relevant for sg events @xcite . we find good agreement with the sk estimate of the average scattering angle as a function of energy ( as reported in @xcite , p. 99 ) . concerning the neutrino fluxes , we refer to @xcite except for sg events , where we use the differential spectra from @xcite with geomagnetic corrections @xcite . since the distributions of parent neutrinos in @xcite are given in arbitrary units , we need to normalize the total area of our estimated sk lepton distributions ( sg and mg , in the absence of oscillations ) to the corresponding values simulated by the sk collaboration , as reported in table i ( total rates ) . for sg events , this renormalization compensates , in part , for the fact that we use low - energy @xmath7 fluxes from @xcite instead than from @xcite . of course , a more direct calculation of the sg and mg distributions ( avoiding the use of indirect information such as the parent @xmath7 distributions ) is preferable ; we intend to perform such calculation when the sk efficiency function @xmath309 will be made publicly available . in any case , our present approach produces results in satisfactory agreement with sk zenith distributions for sg and mg events @xcite , as shown in fig . 20 . the small differences between our calculations and the sk simulations are not relevant , being comparable to the sk montecarlo statistical error . figure 20 also shows the up@xmath0 distribution , for which we use a direct computation as in @xcite , with the following ingredients : grv94 dis structure functions @xcite , lohmann _ et al . _ muon energy losses in the rock @xcite , and the zenith dependence of the sk muon energy threshold from @xcite . also for this distribution , we obtain a good agreement with the corresponding sk calculation ( with the same inputs ) . notice that fig . 20 refers to the no oscillation case . some `` oscillated '' @xmath0-like and @xmath1-like event distributions ( as well as their ratio @xmath26 ) have also been presented by the sk collaboration in various conferences , especially for the case of maximal @xmath2 mixing . we obtain good agreement with sk also in such cases ( not shown ) . in conclusion , we are confident that our calculations of the zenith distributions represent a satisfactory approximation ( not a substitute , of course ) of the sk simulations . improvements of our calculations for the sg amd mg samples are possible ( with a more accurate knowledge of the sk detector efficiency ) but do not appear to be decisive at present , in view of the good agreement reported in fig . 20 and of the relatively large theoretical uncertainties discussed in the following appendix . fitting histograms is a delicate task . in general , the predictions in any two bins are correlated , and ignoring such correlations typically leads to significant variations in the allowed ranges for the fit parameters . concerning the sk zenith distributions , this problem adds to the difficulty of evaluating the size of uncertainties associated to the theoretical neutrino flux calculations and to the neutrino cross section , as well as the variations of such uncertainties in terms of the neutrino energy and direction . there is no easy solution to such problems , other than continually improving the calculations , understanding the role and the uncertainties of any input parameter , and removing as many approximations as possible @xcite . cosmic ray experiments can also help to constrain the models of atmospheric showers . the confidence in the estimated cross sections might benefit from a resurgence of experimental interest for low - energy neutrino interactions . in the meantime , it is wise to adopt conservative error estimates . for each of the 30 sk zenith bins @xmath310 used in the analysis , we take the experimental statistical error @xmath311 from tables i and ii . the @xmath312 statistical error matrix @xmath313 simply reads @xmath314 . as a global systematic error @xmath315 for each bin , we assume conservatively @xmath39 of the theoretical prediction ( with or without oscillations ) . the systematic error matrix is then @xmath316 , where the correlation matrix @xmath317 is evaluated as follows . if the systematic uncertainties @xmath315 had a single , common origin such as an overall normalization uncertainty , then the bin values @xmath318 s would be fully correlated ( @xmath319 ) and the systematics would cancel in any bin ratio @xmath320 . however , the presence of several sources of uncertainties implies that @xmath321 and that the ratio @xmath320 is affected by a residual uncertainty @xmath322 where all @xmath302 s represent fractional errors . for @xmath323 the above relation can be inverted to give @xmath324 which allows to estimate @xmath317 from the ratio error . for instance , if @xmath318 refers to downgoing sg @xmath1-like events and @xmath325 to the corresponding @xmath0-like events , with a @xmath26 uncertainty of , say , @xmath25 , the corresponding correlation index is @xmath326 @xcite . the task is then reduced to the evaluation of the most important sources of errors for the ratios @xmath320 . the total error for the @xmath26 flavor ratio ( including the theoretical uncertainties and the experimental misidentification ) is conservatively estimated to be @xmath327 for sg events and @xmath328 for mg events in @xcite . for bins of equal flavor , one expects an additional energy - dependent uncertainty in the ratio @xmath320 due to uncertainties in the neutrino energy spectrum slope . in fact , by comparing the relative rates of sg , mg , and up@xmath0 events calculated with different input fluxes ( either @xcite or @xcite ) , we find typical ratio errors of @xmath25 for @xmath329 and @xmath330 , and of @xmath331 for @xmath332 , i.e. , errors increasing with the relative difference between the mean energies of the event samples . finally , one expects also angular - dependent errors for @xmath320 , that we estimate to be at most @xmath25 when the difference between @xmath333 and @xmath334 is maximal ( ratio of vertical to horizontal direction bins ) . qualitatively , all this means that the correlation between any two bins decreases from unity as the bins are more separated in energy , angle , and flavor , thus giving to the theoretical distributions some freedom to vary their shape . quantitatively , we formalize the above estimates by generalizing eq . ( [ rho ] ) as @xmath335 where @xmath336 , and : ( i ) @xmath337 is the `` flavor - dependent uncertainty , '' equal to 10% for bins of different flavors and zero otherwise ; ( ii ) @xmath338 is the `` energy - dependent uncertainty , '' equal to zero for bins @xmath339 belonging to the same sample ( sg , mg , or up ) , to 5% for bins @xmath339 of the kind ( sg , mg ) or ( mg , up ) , and to 10% for bins @xmath339 of the kind ( sg , up ) ; and ( iii ) @xmath340 is the `` direction - dependent uncertainty , '' equal to @xmath341 times the difference between the mean direction cosines @xmath342 and @xmath343 . for instance , the first bin of the sg@xmath1 distribution and the last bin of the up@xmath0 distribution have the lowest correlation , @xmath344 , since they are the most distant in energy , flavor , and direction . we finally define our @xmath63 function as @xmath345 where @xmath346 is the difference between the bin contents in tables i and ii and our theoretical calculations ( with or without oscillations ) . we mention that the @xmath63 fit to the sk data appears to be rather sensitive to @xmath337 . lowering its value from our present choice ( 10% , comparable to the estimates in table ii of @xcite ) to a few percent would shrink significantly the allowed regions but would also worsen the best fit . a reduction of this and other systematics would greatly improve the statistical power of oscillation hypothesis tests . soudan 2 collaboration , w.w.m . allison _ et al . _ , phys . b * 391 * , 491 ( 1997 ) . t. kafka , in _ taup97 _ , proceedings of the 5th international conference on topics in astroparticle and underground physics ( gran sasso , italy , 1997 ) , to appear in nucl . b ( proc . suppl . ) . t. kajita , in _ neutrino 98 _ , proceedings of the xviii international conference on neutrino physics and astrophysics , takayama , japan , june 1998 , to appear . transparencies available at the url http://www-sk.icrr.u-tokyo.ac.jp/nu98/scan/ . e. lisi , proceedings of _ new era in neutrino physics _ , a satellite symposium after _ neutrino 98 _ @xcite , tokyo metropolitan university , tokyo , june 1998 , to appear . transparencies available at the url http://musashi.phys.metro-u.ac.jp/neutrino.html . mass2 collaboration , p. papini _ et al . _ , in _ icrc95 _ , proceedings of the 24th international conference on cosmic rays , rome , italy , 1995 , edited by h. moraal , j.k . gaisser , and j. arafune ( arti grafiche editoriali , urbino , 1995 ) , vol . 4 , p. 1033 ; g. basini _ et al . _ , _ ibidem _ , vol . 1 , p. 585 . mass2 collaboration , g. basini _ et al . _ , in _ icrc97 _ , proceedings of the 25th international conference on cosmic rays , durban , south africa , edited by m.s . potgieter , b.c . raubenheimer , and d.j . van der walt ( potchefsroomse universiteit , 1997 ) , vol . 6 , p. 381 . gaisser and t. stanev , in _ icrc95 _ , proceedings of the 24th international conference on cosmic rays , rome , italy , 1995 , edited by h. moraal , j.k . gaisser , and j. arafune ( arti grafiche editoriali , urbino , 1995 ) , vol . 1 , p. 694	we present a detailed analysis of the zenith angle distributions of atmospheric neutrino events observed in the super - kamiokande ( sk ) underground experiment , assuming two - flavor and three - flavor oscillations ( with one dominant mass scale ) among active neutrinos . in particular , we calculate the five angular distributions associated to sub - gev and multi - gev @xmath0-like and @xmath1-like events and to upward through - going muons , for a total of 30 accurately computed observables ( zenith bins ) . first we study how such observables vary with the oscillation parameters , and then we perform a fit to the experimental data as measured in sk for an exposure of 33 kty ( 535 days ) . in the two - flavor mixing case , we confirm the results of the sk collaboration analysis , namely , that @xmath2 oscillations are preferred over @xmath3 , and that the no oscillation case is excluded with high confidence . in the three - flavor mixing case , we perform our analysis with and without the additional constraints imposed by the chooz reactor experiment . in both cases , the analysis favors a dominance of the @xmath2 channel . without the chooz constraints , the amplitudes of the subdominant @xmath3 and @xmath4 transitions can also be relatively large , indicating that , at present , current sk data do not exclude sizable @xmath5 mixing by themselves . after combining the chooz and sk data , the amplitudes of the subdominant transitions are constrained to be smaller , but they can still play a nonnegligible role both in atmospheric and other neutrino oscillation searches . in particular , we find that the @xmath5 appearance probability expected in long baseline experiments can reach the testable level of @xmath6 . we also discuss earth matter effects , theoretical uncertainties , and various aspects of the statistical analysis .
the basis of stochastic superspace models is the assumption that the grassmannian coordinates , @xmath5 and @xmath6 , are stochastic variables . writing down the most general probability distribution consistent with lorentz invariance , we find that only one additional parameter is required such that @xcite , @xmath7 where @xmath2 , a complex number of mass dimension , is the measure of stochasticity . the lagrangian in ordinary spacetime is then found by averaging the supersymmetric lagrangian over the probability distribution in eq . . applying this procedure to the superpotential of the minimal supersymmetric standard model , @xmath8 leads to the soft - breaking terms @xmath9 + \text{h.c.},\ ] ] where tildes represent the scalar component of the chiral superfield . similarly , averaging over the gauge - kinetic @xmath10 densities results in soft - breaking masses for the gauginos , @xmath11 , @xmath12_f - \frac{\xi^}{2 } \sum_i \text{tr } \lambda^{\left(i\right ) } \lambda^{\left(i\right ) } + \text{h.c.},\ ] ] where @xmath13 denotes the standard model field - strength superfields . see @xcite for further details . referring to eqns . and , it is clear that the tree - level soft - breaking terms for minimal stochastic superspace are @xcite : a. the bilinear higgs soft term , @xmath14 , b. the universal trilinear scalar soft terms proportional to the yukawa couplings , @xmath15 , c. the universal gaugino masses , @xmath16 , d. the universal soft scalar masses , @xmath17 . as these soft - breaking terms are renormalised at the quantum level , the relations listed above are only defined as such at some energy cutoff scale , @xmath18 , which becomes an additional parameter of the theory . the emergence of a pattern of soft - breaking terms dependent only on the parameter @xmath2 categorises stochastic superspace models as a special case of the broader constrained supersymmetric standard model ( cmssm ) . although models with absent soft scalar masses are typically considered excluded at 95@xmath19 c.l . @xcite ] , this is because the soft - breaking parameters are customarily defined at @xmath20 . we have found that for larger cutoff scales , @xmath21 , the neutralino becomes the lsp , and consequently , one uncovers a new phenomenologically viable region of parameter space @xcite . we look at this in further detail in sec . [ sec : scan ] by computing the neutralino relic density . we have previously demonstrated the existence of phenomenologically viable regions of the @xmath22 parameter space falling within the stau coannihilation regime @xcite . in this study we expand the scope of the analysis primarily to examine a broader range of @xmath2 values and include a calculation of the cold dark matter relic density . most publicly available sparticle spectrum software packages take the msugra parameters @xmath23 , @xmath24 , @xmath25 , @xmath26 and @xmath27 as inputs at the cutoff scale . however , stochastic superspace makes a prediction for the bilinear higgs soft term @xmath28 , rather than explicitly predicting a value for @xmath26 . to facilitate a more thorough analysis of the model , we previously modified the program softsusy 3.1.7 @xcite to swap @xmath26 for @xmath28 as an input . as branching ratios are required for this analysis , we use the package isajet 7.81 @xcite to calculate the sparticle and decay spectra , using the value of @xmath26 found with softsusy . various packages included in isatools calculate the relic density and branching ratios of rare decays . we scan the parameter regions @xmath29 over 150 points in @xmath2 and 50 points in @xmath18 for a total of 7500 points , against the following conditions : @xmath30 in stochastic superspace , the character of the lsp is dependent on one s parameter choice , and will either be the lightest neutralino or the lightest stau . since r - parity is assumed to be conserved , we enforce the condition in eq . to ensure all instances of a stau lsp are excluded . the lightest @xmath31even higgs must satisfy condition , where we take into consideration the theoretical error of @xmath32 @xcite in the published limit of @xmath33 at 95% cl @xcite . conditions - arise from supersymmetric particle searches @xcite . condition , the upper limit on the branching fraction of the process @xmath1 is as measured by the cms and lhcb collaborations @xcite . finally , condition shows the acceptable range of relic density based on the wmap 7-year mean , @xmath34 @xcite , and the theoretical error of @xmath35 @xcite , shown with errors added in quadrature . applying all constraints listed in eqns . - , we find a phenomenologically viable region of parameter space , shown in black in fig . [ fig : rd ] . relaxing condition , we find another region of interest where @xmath36 ( see gray region in fig . [ fig : rd ] ) . although the relic density is lower than observed , this region is not necessarily excluded as the minimal model of stochastic superspace may be extended to include an additional source of dark matter . furthermore , theoretical uncertainties in the determination of the sparticle mass spectrum can have a significant effect on the calculated value of the relic density , where a 1% difference in mass could affect the calculated relic density by up to 10% @xcite . it is possible , therefore , that the region of parameter space satisfying the relic density experimental bounds may be larger than depicted in fig [ fig : rd ] . the region where @xmath37 and/or @xmath38 is excluded due to sparticle masses violating conditions - or going tachyonic . excluded points in the region where @xmath39 and @xmath40 have relic density larger than the upper bound in condition . in the region with @xmath41 and @xmath40 , a number of points are excluded for having a stau lsp . for clarity , figure [ fig : rdexcl ] displays these regions graphically . all points in the parameter space satisfying constraints - above , with relic density within or below the bound given in condition , form the phenomenologically viable region of stochastic superspace . the atlas and cms experiments @xcite have recently updated their searches for supersymmetric particles using the 2011 dataset @xcite . data collected from proton collisions at the large hadron collider at @xmath42 tev are analysed in a variety of final states . no excess over the standard model expectation has been observed , allowing one to set exclusion limits in a variety of candidate model spaces such as the cmssm , the model that most closely resembles that considered here . as the lhc is a proton - proton collider , one expects to dominantly produce coloured objects such as squarks and gluinos , whose inclusive leptonic branching ratios are relatively small , and hence the strongest cmssm exclusions to date result from the atlas searches for events with no leptons and the cms searches for sparticle production in hadronic final states . the atlas and cms limits have a similar reach in the squark and gluino masses , and here we consider only the atlas zero lepton limits for simplicity . recent interpretations of lhc limit results can be found in @xcite . atlas defined a series of signal regions in which to look for sparticle production , each tuned to enhance sensitivity in a particular region of the @xmath23-@xmath24 plane . events with an electron or muon with @xmath43 20 gev were rejected . table [ tab : cuts ] summarises the remaining selection cuts for each region , whilst table [ tab : results ] gives the observed and expected numbers of events . these numbers were used by the atlas collaboration to derive limits on @xmath44 , where @xmath45 is the cross - section for new physics processes for which the atlas detector has an acceptance @xmath46 and a detector efficiency of @xmath47 . these results are also quoted in table [ tab : results ] . .[tab : cuts ] selection cuts for the five atlas zero lepton signal regions . @xmath48jet , @xmath49 is the smallest of the azimuthal separations between the missing momentum @xmath50 and the momenta of jets with @xmath51 40 gev ( up to a maximum of three in descending @xmath52 order ) . the effective mass @xmath53 is the scalar sum of @xmath54 and the magnitudes of the transverse momenta of the two , three and four highest @xmath52 jets depending on the signal region . in the region rhm , all jets with @xmath5140 gev are used to define @xmath53 . [ cols="^,^,^,^,^,^",options="header " , ] the atlas collaboration has used the absence of evidence of sparticle production in 1 fb@xmath0 of data to place an exclusion limit at the 95% confidence level in the @xmath23-@xmath24 plane of the cmssm for fixed @xmath25 and tan@xmath55 , and for @xmath56 . although the stochastic susy model considered here may be considered a subset of the cmssm , it is non - trivial to recast this limit into a constraint on the parameters @xmath2 and @xmath18 . given a signal expectation for a particular model , however , one can easily evaluate the likelihood of that model using the published atlas background expectation and observed event yield in each search channel . by simulating points in the @xmath2-@xmath18 plane , we can therefore investigate the lhc exclusion reach in stochastic susy space , provided that we can demonstrate that our simulation provides an adequate description of the atlas detector . in this paper , we use isajet 7.81 @xcite to produce susy mass and decay spectra then use herwig++ 2.5.1 @xcite to generate 10,000 monte carlo events . delphes 1.9 @xcite is subsequently used to provide a fast simulation of the atlas detector . the total susy production cross - section is calculated at next - to - leading order using prospino 2.1 @xcite , where we include all processes except direct production of neutralinos , charginos and sleptons since the latter are subdominant . the atlas setup differs from this only in the final step of detector simulation , where a full , geant 4 based simulation @xcite is used to provide a very detailed description of particle interactions in the atlas detector at vast computational expense . it is clear that the delphes simulation will not reproduce every result of the advanced simulation . nevertheless , one can assess the adequacy of our approximate results by trying to reproduce the atlas cmssm exclusion limits . we have generated a grid of points in the @xmath23-@xmath24 plane using the same fixed values of tan@xmath55=10 and @xmath25=0 as the published atlas result . we must now choose a procedure to approximate the atlas limit setting procedure . atlas use both @xmath57 and profile likelihood methods to obtain a 95% confidence limit , using a full knowledge of the systematic errors on signal and background . although the systematic error on the background is provided in the atlas paper , we do not have full knowledge of the systematics on the signal expectation , which may in general vary from point to point in the @xmath23-@xmath24 plane . rather than implement these statistical techniques , we take a similar approach to that used in @xcite , and use the published @xmath44 limits to determine whether a given model point is excluded in a search channel . we use our simulation to obtain the @xmath44 value for a given model point , and consider the model to be excluded if the value lies above the limit given in table [ tab : results ] . this allows us to draw an exclusion contour in each search channel , and we estimate the combined limit by taking the union of the individual exclusion contours for each channel ( i.e. the most stringent search channel for a given model is used to determine whether it is excluded ) . the procedure defined above neglects systematic errors on the signal and background yields and , as noted in @xcite , this leads to a discrepancy between the delphes results and the atlas limits in each channel . we follow @xcite in using a channel dependent scaling to tune the delphes output so that the limits in each channel match as closely as possible `` by eye '' . we obtain factors of 0.82 , 0.85 , 1.25 , 1.0 and 0.70 for the r1 , r2 , r3 , r4 and rhm regions respectively . comparisons between the resulting delphes exclusion limit and the atlas limit are shown in figure [ fig : limits1 ] , where we observe generally good agreement in all channels . the largest discrepancy is observed in the rhm channel , where we find that one can not get the tail of the limit at large @xmath23 to agree with the atlas limit whilst simultaneously guaranteeing good agreement at low @xmath23 . this is likely to be due to the fact that we have effectively assumed a flat systematic error over the @xmath58 plane . whereas the atlas results use a full calculation of the systematic errors for each signal point . it is important to notice however that the _ combined _ limit will be dominated by regions r1 and r2 at low @xmath23 , and thus by choosing to tune the rhm results in order to reproduce the large @xmath23 tail , one can ensure reasonable agreement of the combined limit over the entire range . where disagreement remains , the delphes limit is less stringent than the atlas limit , and hence using it gives us a conservative estimate of the atlas exclusion reach . ( 8.7,7.1)(-0.7,-0.1 ) ( -0.5,4.8 ) ( 2.7,4.8 ) ( -0.5,2.3 ) ( 2.7,2.3 ) ( -0.5,-0.2 ) ( 2.7,-0.2 ) having obtained reasonable agreement with the atlas cmssm limits in the previous section , we now turn our attention to calculating the lhc exclusion zone in the plane of stochastic susy parameters @xmath2 and @xmath18 . to avoid unnecessary ( and computationally expensive ) simulation , we only simulate those points that passed the constraints detailed in sec . [ sec : scan ] . ( 5.8,5.2)(-0.15,0 ) ( 0,0 ) ( 2.7,0 ) ( 0,2.6 ) ( 2.7,2.6 ) values for stochastic susy points for atlas signal region rhm . the atlas limit for this signal region is 17 fb , which excludes stochastic superspace points with @xmath59 in the region to the right of the light green colour . ] figures [ fig : colz1 ] and [ fig : colz2 ] show the @xmath44 values ( including the systematic factors ) for each of the atlas search channels . @xmath44 is most strongly dependent on @xmath2 which is to be expected ; points at low @xmath60 have light coloured sparticles and will thus have large production cross - sections at the lhc . raising the mass scale essentially reduces the production cross - section leading to a corresponding decrease in the @xmath44 value as one travels left along the @xmath2 axis . there is little to no dependence of @xmath44 on the cutoff scale for a given value of @xmath2 . we can conservatively bound the region @xmath61 as consistent with 1 fb@xmath0 of lhc data based on atlas signal region r1 , which is the most constraining in the case of stochastic superspace . in this paper , we perform a systematic scan of the parameter space of the stochastic superspace model of soft supersymmetry breaking . its phenomenology is checked against constraints from direct sparticle searches , cold dark matter relic density and the branching ratio of the process @xmath1 . points that pass these conditions are analysed further , obtaining a value for @xmath44 , which can be directly compared to atlas search limits from the zero lepton channels . we find that the stochasticity parameter is restricted to the region @xmath61 , while the viable region of the cutoff scale , @xmath18 , shrinks with increased @xmath62 , bounded above by relic density concerns , and below from sparticle mass constraints . since a prime motivation for the the low - energy supersymmetry is a solution to the hierarchy problem , the increased bounds on sparticle masses seriously undermine its validity . the required tuning of parameters is estimated at the level of @xmath63 for @xmath64 @xcite . thus , stochastic supersymmetry still allows a region of parameters that provide a satisfactory solution to the hierarchy problem . however , as has been shown in @xcite , the mass of the lightest standard model - like higgs boson is predicted in a very narrow range @xmath65 in the minimal model . this range is outside the @xmath66 region for the higgs mass hinted in the very recent atlas @xcite and cms @xcite data . if the latest evidence for the higgs mass will be confirmed , the minimal model with stochastic supersymmetry will certainly be excluded , motivating research into a non - minimal implementation of stochastic superspace consistent with these results . s. p. martin , in kane , g.l . ( ed . ) : perspectives on supersymmetry ii * 1 - 153 [ hep - ph/9709356 ] . c. f. kolda , nucl . suppl . * 62 * , 266 ( 1998 ) [ hep - ph/9707450 ] . a. kobakhidze , n. pesor and r. r. volkas , phys . d * 79 * , 075022 ( 2009 ) [ arxiv:0809.2426 [ hep - ph ] ] . t. aaltonen _ et al . _ [ cdf collaboration ] , phys . rev . lett . * 103 * , 021802 ( 2009 ) [ arxiv:0902.1266 [ hep - ex ] ] . a. kobakhidze , n. pesor and r. r. volkas , phys . d * 81 * , 095019 ( 2010 ) [ arxiv:1003.4782 [ hep - ph ] ] . b. c. allanach , comput . commun . * 143 * , 305 ( 2002 ) [ hep - ph/0104145 ] . f. e. paige , s. d. protopopescu , h. baer and x. tata , hep - ph/0312045 . b. c. allanach , a. djouadi , j. l. kneur , w. porod and p. slavich , jhep * 0409 * , 044 ( 2004 ) [ hep - ph/0406166 ] . k. nakamura _ et al . _ [ particle data group collaboration ] , j. phys . g g * 37 * , 075021 ( 2010 ) . s. chatrchyan _ et al . _ [ cms collaboration ] , arxiv:1107.5834 [ hep - ex ] . j. serrano , arxiv:1111.2620 [ hep - ex ] . e. komatsu _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] . o. buchmueller , r. cavanaugh , a. de roeck , m. j. dolan , j. r. ellis , h. flacher , s. heinemeyer and g. isidori _ et al . _ , arxiv:1110.3568 [ hep - ph ] . g. belanger , s. kraml and a. pukhov , phys . d * 72 * , 015003 ( 2005 ) [ hep - ph/0502079 ] . b. c. allanach , g. belanger , f. boudjema and a. pukhov , jhep * 0412 * , 020 ( 2004 ) [ hep - ph/0410091 ] . g. aad _ et al . _ [ atlas collaboration ] , jinst * 3 * , s08003 ( 2008 ) . r. adolphi _ et al . _ [ cms collaboration ] , jinst * 3 * , s08004 ( 2008 ) . g. aad _ et al . _ [ atlas collaboration ] , arxiv:1109.6572 [ hep - ex ] . g. aad _ et al . _ [ atlas collaboration ] , arxiv:1111.4116 [ hep - ex ] . g. aad _ et al . _ [ atlas collaboration ] , arxiv:1110.6189 [ hep - ex ] . g. aad _ et al . _ [ atlas collaboration ] , jhep * 1111 * , 099 ( 2011 ) [ arxiv:1110.2299 [ hep - ex ] ] . g. aad _ et al . _ [ atlas collaboration ] , arxiv:1109.6606 [ hep - ex ] . s. chatrchyan _ et al . _ [ cms collaboration ] , arxiv:1109.2352 [ hep - ex ] . [ cms collaboration ] , cms - pas - sus-11 - 005 . [ cms collaboration ] , cms - pas - sus-11 - 004 . o. buchmueller , r. cavanaugh , d. colling , a. de roeck , m. j. dolan , j. r. ellis , h. flacher and s. heinemeyer _ et al . _ , eur . j. c * 71 * , 1634 ( 2011 ) [ arxiv:1102.4585 [ hep - ph ] ] . n. desai and b. mukhopadhyaya , arxiv:1111.2830 [ hep - ph ] . c. beskidt , w. de boer , d. i. kazakov , f. ratnikov , e. ziebarth and v. zhukov , phys . b * 705 * , 493 ( 2011 ) [ arxiv:1109.6775 [ hep - ex ] ] . o. buchmueller , r. cavanaugh , d. colling , a. de roeck , m. j. dolan , j. r. ellis , h. flacher and s. heinemeyer _ et al . _ , eur . j. c * 71 * , 1722 ( 2011 ) [ arxiv:1106.2529 [ hep - ph ] ] . b. c. allanach , t. j. khoo and k. sakurai , arxiv:1110.1119 [ hep - ph ] . s. gieseke , d. grellscheid , k. hamilton , a. papaefstathiou , s. platzer , p. richardson , c. a. rohr and p. ruzicka _ et al . _ , arxiv:1102.1672 [ hep - ph ] . s. ovyn , x. rouby and v. lemaitre , arxiv:0903.2225 [ hep - ph ] . w. beenakker , r. hopker and m. spira , hep - ph/9611232 . s. agostinelli _ et al . _ [ geant4 collaboration ] , nucl . instrum . a * 506 * , 250 ( 2003 ) . [ atlas collaboration ] , atlas - conf-2011 - 163 . [ cms collaboration ] , cms - pas - hig-11 - 032 .	we analyse restrictions on the stochastic superspace parameter space arising from 1 fb@xmath0 of lhc data , and bounds on sparticle masses , cold dark matter relic density and the branching ratio of the process @xmath1 . a region of parameter space consistent with these limits is found where the stochasticity parameter , @xmath2 , takes values in the range @xmath3 , provided the cutoff scale is @xmath4 . supersymmetry is a popular extension to the standard model , renowned for its solution to the hierarchy problem , while also providing a dynamical mechanism of electroweak symmetry breaking and a dark matter candidate in its minimal form ( see @xcite for a review ) . by postulating a symmetry relating fermionic and bosonic states , it predicts the existence of a superpartner of opposite spin statistics and degenerate mass for each standard model particle . these superpartners have not been observed , requiring that supersymmetry be softly broken for it to remain a viable model . a number of methods for breaking supersymmetry exist , which can be broadly categorised into two groups : those originating from a fundamental theory where supersymmetry is spontaneously broken in a hidden sector , then communicated to the visible sector by a messenger , and the purely phenomenological approach , where no explanation is provided as to the origins of supersymmetry breaking . although fundamental theories provide a dynamical explanation for the mechanism of supersymmetry breaking , complications such as large sparticle induced fcnc amplitudes in gravity mediation , and cp violation issues in gauge mediation , leave no single compelling solution @xcite . alternatively , the phenomenological approach of writing down the most general lagrangian with explicit soft - breaking terms fails to address the source of susy breaking , while introducing an overwhelmingly vast new parameter space . stochastic superspace is a unique mechanism for softly breaking supersymmetry that bridges the two approaches discussed above @xcite . by considering the grassmannian coordinates to be stochastic variables , a very constrained set of soft - breaking terms emerges . though the underlying cause of stochasticity is not postulated , it introduces predictability not achievable through the purely phenomenological approach . in this paper we investigate the phenomenology of stochastic superspace at the lhc . for each suitable point in the stochastic superspace parameter space , we use a fast simulation of the atlas detector to investigate whether points are excluded by the atlas zero lepton searches , which in turn are amongst the most constraining current limits on direct sparticle production . the rest of the paper is structured as follows : sec . [ sec : overview ] provides an overview of stochastic superspace , sec . [ sec : scan ] discusses constraints applied in a scan of its parameter space , and sec . [ sec : lhc ] describes the technique used to simulate stochastic superspace model points at the atlas detector , and explains how we approximate the atlas exclusion limits on supersymmetric particle production . we terminate with concluding remarks in sec . [ sec : conclusion ] .
several papers of the first author and borisov were devoted to a combinatorial method for explicit constructions of mirror pairs of calabi - yau varieties ( see @xcite ) . a starting point of this method was the notion of a _ reflexive polytope _ introduced in @xcite . a reflexive polytope @xmath8 is a convex @xmath1-dimensional polytope with vertices in @xmath9 containing the origin @xmath10 in its interior such that its dual ( polar ) polytope @xmath11 is again a lattice polytope ( i.e. , all vertices of @xmath4 belong to @xmath9 ) . in this case @xmath4 is also a reflexive polytope and @xmath12 . for any convex lattice polytope @xmath13 we denote by @xmath14 a generic hypersurface in @xmath15 defined by the equation @xmath16 where the coefficients @xmath17 @xmath18 are independent variables and @xmath19 are standard complex coordinates on @xmath15 . if @xmath3 is reflexive , then @xmath20 admits a natural compactification @xmath21 which is a @xmath22-dimensional calabi - yau variety . moreover , the duality between @xmath3 and @xmath4 becomes the mirror duality between @xmath22-dimensional calabi - yau varieties @xmath21 and @xmath23 many explicit constructions of mirrors of calabi - yau complete intersections suggested by the first author and van straten in @xcite motivated borisov to generalize the polar duality for reflexive polytopes to a more general duality for so called _ nef - partitions _ ( see @xcite ) . from a combinatorial point of view , a _ nef - partition of length @xmath0 _ is a decomposition @xmath24 of a @xmath1-dimensional reflexive polytope @xmath25 into a minkowski sum of @xmath0 lattice polytopes @xmath26 such that @xmath27 for all @xmath28 ( it is not assumed that each @xmath29 has the maximal dimension @xmath1 ) . it can be shown that for any @xmath30 the polytope @xmath31 has vertices in @xmath32 , @xmath33 and the minkowski sum @xmath34 is again a @xmath1-dimensional reflexive polytope . moreover , the lattice polytopes @xmath26 can be obtained from lattice polytopes @xmath35 by the same formula : @xmath36 the minkowski sum decomposition @xmath34 is called the _ dual nef - partition_. some examples of nef - partitions and their duals are given in examples [ nefp ] , [ dual - nefp ] and [ nef - p2 ] . if @xmath37 is a generic complete intersection in @xmath15 defined by generic equations @xmath38 then @xmath37 admits a natural compactification @xmath39 which is a @xmath40-dimensional calabi - yau variety . moreover , the duality of nef - partitions @xmath41 and @xmath42 becomes the mirror duality between @xmath40-dimensional calabi - yau varieties @xmath39 and @xmath43 . if @xmath25 is a reflexive polytope , @xmath44 is a nef - partition and @xmath45 is the dual nef - partition , then the lattice polytopes @xmath46 satisfy some additional combinatorial conditions : * @xmath47 for all @xmath48 ; * the convex hull @xmath49 is the dual reflexive polytope @xmath50 ; * @xmath51 . moreover , @xmath52 is a vertex of @xmath50 if and only if @xmath53 is a nonzero vertex of @xmath29 for some @xmath54 . for lattice polytopes @xmath55 we denote by @xmath56 the polytope in @xmath57 which is the convex hull @xmath58 where @xmath59 is the standard basis of @xmath60 . we call @xmath56 _ cayley polytope associated with _ @xmath61 . let @xmath3 and @xmath62 be two @xmath1-dimensional reflexive polytopes and @xmath63 and @xmath34 nef - partitions which are dual to each other . we consider two @xmath64-dimensional polytopes @xmath65 , @xmath66 and two @xmath67-dimensional cones @xmath68 @xmath69 it was proved in ( * ? ? ? * theorem 4.6 ) that the cones @xmath70 and @xmath71 are dual to each other with respect to standard scalar product on @xmath57 . moreover , there exists a unique lattice point @xmath72 ( resp . @xmath73 ) in the relative interior of @xmath74 ( resp . @xmath75 ) such that @xmath76 ( resp . @xmath77 ) is a @xmath64-dimensional reflexive polytope . for this reason , the duality for nef - partitions of length @xmath0 can be considered as a special case of the more general duality for so called _ reflexive gorenstein cones of index @xmath0_. the duality for reflexive gorenstein cones gives new possibilities for constructing mirrors of calabi - yau varieties . in particular , it allows to construct mirrors of rigid - calabi - yau @xmath78-folds in form of so called _ generalized calabi - yau varieties _ or _ landau - ginzburg orbifolds _ @xcite . recall that a @xmath64-dimensional lattice polytope @xmath79 is called a _ gorenstein polytope of index @xmath0 _ , if @xmath80 contains a single interior lattice point @xmath81 and @xmath82 is a reflexive polytope . in this case , the @xmath67-dimensional cone @xmath83 is called _ reflexive gorenstein cone of index @xmath0 _ associated with @xmath3 . let @xmath84{}^{{\scriptstyle\vee}}}:= \ { y \in \r^d \times \r , \ ; \sum_k { x_k}{y_k } \geq 0 , \;\forall x \in c_p \}\ ] ] be the dual cone . then there exists another gorenstein polytope @xmath4 of index @xmath0 such that @xmath85 isomorphic ( via the action of @xmath86 ) to the cone @xmath87{}^{{\scriptstyle\vee}}}$ ] . the polytope @xmath4 is uniquely determined up to the action of the affine group @xmath88 on @xmath89 . it is called the _ dual gorenstein polytope of index @xmath0_. in this paper , we want to look at the duality of reflexive gorenstein cones of index @xmath0 from the viewpoint of the duality of gorenstein polytopes of index @xmath0 . for @xmath2 the latter coincides with the already known polar duality for reflexive polytopes . it is important to consider an additional structure on gorenstein polytopes @xmath3 which is defined by a choice of a special @xmath5-dimensional lattice simplex @xmath6 . the combinatorial notion of a _ special simplex _ was recently introduced in @xcite in order to prove a conjecture of stanley . if a gorenstein polytope @xmath3 of index @xmath0 and its dual @xmath4 both contain special @xmath5-simplices , we natually obtain two nef - partitions dual to each other . moreover , the existence of special simplices in @xmath3 and @xmath4 can be considered as another characterization of nef - partitions and their duals ( see proposition [ splittheo ] ) . the choice of special @xmath5-simplices is not unique . therefore , one may obtain many different nef - partitions from the same pair of gorenstein polytopes @xmath90 . we expect that different choices of special @xmath5-simplices @xmath6 and @xmath7 define calabi - yau complete intersections which are equivalent from the viewpoint of homological mirror symmetry . the paper is organized as follows . in section 1 we recall the notions of gorenstein polytopes , reflexive gorenstein cones and their dualities . in section 2 we establish a bijective correspondence between cayley structures of length @xmath0 on a gorenstein polytope @xmath3 of index @xmath0 and special @xmath5-simplices in the dual gorenstein polytope @xmath4 . in section 3 we review the notion of nef - partitions and their duality as well as combinatorial properties of nef - partitions . in section 4 we define the @xmath91-function of a gorenstein polytope @xmath3 and formulate some open questions and conjectures . finally , in section 5 we give a precise formulation of a conjecture which connects special @xmath5-simplices in a gorenstein polytope @xmath3 with the homological mirror symmetry . in the last section 6 we consider as an addendum to section 3 some combinatorial operations with nef - partitions . we are grateful to anonymous referees for valuable remarks . the second author thanks christian haase for discussions , motivating questions and suggestions . let @xmath92 be a lattice of rank @xmath1 , @xmath93 the dual lattice , @xmath94 the natural pairing . we define @xmath95 , @xmath96 , @xmath97 , @xmath98 . by a polytope @xmath99 we always mean a convex hull of finitely many points in @xmath100 . the set of vertices of @xmath101 is denoted by @xmath102 , the relative interior by @xmath103 , and the boundary by @xmath104 . for arbitrary subset @xmath105 we denote by @xmath106 ( resp . @xmath107 ) the convex ( resp . affine ) hull of @xmath108 . from now on we assume @xmath109 to be a @xmath1-dimensional _ lattice polytope _ , i.e. , @xmath110 . assume that @xmath101 has the origin @xmath111 in its interior . then the _ dual polytope _ of @xmath101 is defined as @xmath112 this is a rational polytope ( i.e. , @xmath113 ) containing the origin @xmath114 in its interior . if @xmath101 contains an interior lattice point @xmath115 ( not necessary @xmath116 ) , we say that @xmath117 is the _ dual polytope of @xmath101 with respect to @xmath53_. let @xmath109 be a lattice polytope containing the origin @xmath111 in its interior . we say @xmath101 is _ reflexive _ , if the dual polytope @xmath118 is also a lattice polytope . in this case , @xmath119 is a single interior lattice point . in general , if a lattice polytope @xmath99 contains a single interior lattice point @xmath53 ( not necessary @xmath116 ) , we say that @xmath101 is _ reflexive with respect to @xmath53 _ , if @xmath120 is reflexive . it is easy to see that a lattice polytope @xmath109 is reflexive if and only if every facet of @xmath101 has integral distance @xmath121 from its unique interior lattice point . recall that a point @xmath122 is said to have _ integral distance _ @xmath123 from a facet @xmath124 of @xmath101 , if there is a primitive lattice point @xmath125 such that @xmath126 and @xmath127 . the following example shows that in dimension @xmath128 there exist lattice polytopes containing @xmath116 as a single interior lattice point which are not reflexive . let @xmath101 be the @xmath78-dimensional lattice polytope with vertices @xmath129 , @xmath130 , @xmath131 . then @xmath132 is a single interior lattice point , however , the integral distance from @xmath116 to the facet @xmath133 is @xmath134 , since for the primitive vector @xmath135 we have @xmath126 and @xmath136 . let @xmath0 be a positive integer . a lattice @xmath1-dimensional polytope @xmath99 is called _ gorenstein of index _ @xmath0 , if @xmath137 contains an interior lattice point @xmath53 and @xmath138 is reflexive . let @xmath13 be the standard unit cube : @xmath139 @xmath140 . then @xmath141 is reflexive . therefore @xmath101 is a gorenstein polytope of index @xmath134 . it is clear that gorenstein polytopes can be characterized as follows : a lattice polytope @xmath101 is a gorenstein polytope of index @xmath0 if and only if there is a rational point @xmath142 having the integral distance @xmath143 from any facet of @xmath101 . let @xmath144 and @xmath145 be lattices of rank @xmath146 which are dual to each other . let @xmath147 be a @xmath146-dimensional rational finite polyhedral cone with vertex @xmath148 . then @xmath149{}^{{\scriptstyle\vee}}}:= \{y \in \nrq \,:\ ; { \langle x , y \rangle } \geq 0 \ ; \forall\ , x \in \sigma\}\ ] ] is called the _ dual cone _ of @xmath150 . it is also a @xmath146-dimensional rational finite polyhedral cone with vertex @xmath116 . we recall some definitions from @xcite : a @xmath146-dimensional rational finite polyhedral cone @xmath150 is called _ gorenstein cone _ , if it is generated by finitely many lattice points which are contained in an affine hyperplane @xmath151 for some @xmath152 . in this case , the lattice point @xmath153{}^{{\scriptstyle\vee}}})$ ] is uniquely determined , and we denote it by @xmath154 . one has @xmath155{}^{{\scriptstyle\vee } } } ) \cap \nq = \ns + \sigma^{\makebox[0mm]{}^{{\scriptstyle\vee}}}\cap \nq$ ] . we define the @xmath156th slice of @xmath150 as @xmath157 . [ gor - cone ] let @xmath158 be a gorenstein cone . then the lattice polytope @xmath159 is called the _ support _ of @xmath150 ; it is a @xmath160-dimensional lattice polytope with respect to the affine lattice @xmath161 . let @xmath109 be a lattice polytope of dimension @xmath1 . we set @xmath162 , @xmath163 , @xmath164 , @xmath165 . then @xmath166 is a gorenstein cone with support @xmath167 . using the two constructions above , we obtain a @xmath121-to-@xmath121 correspondence between @xmath146-dimensional gorenstein cones @xmath168 considered up to action of @xmath169 on @xmath144 and @xmath160-dimensional lattice polytopes @xmath170 considered up to an affine transformation from @xmath171 . a gorenstein cone @xmath150 is called _ reflexive _ , if @xmath172{}^{{\scriptstyle\vee}}}$ ] is also a gorenstein cone . in this case , we denote by @xmath173 the unique lattice point in @xmath144 such that the affine hyperplane @xmath174 contains all lattice points generating the cone @xmath172{}^{{\scriptstyle\vee}}}$ ] . the number @xmath175 is called the _ index _ of the reflexive gorenstein cone @xmath150 . using @xmath176{}^{{\scriptstyle\vee}}})^{\makebox[0mm]{}^{{\scriptstyle\vee}}}= \sigma$ ] , we see that @xmath0 is also the index of the reflexive gorenstein cone @xmath172{}^{{\scriptstyle\vee}}}$ ] . thus , we obtain a natural duality of reflexive gorenstein cones of index @xmath0 . [ refl - g ] it was observed in ( * ? ? ? * prop.2.11 ) that reflexive gorenstein cones correspond to gorenstein polytopes : let @xmath150 be a @xmath146-dimensional gorenstein cone with support @xmath167 . then the following statements are equivalent : 1 . @xmath150 is a reflexive gorenstein cone of index @xmath0 ; 2 . @xmath177 is a reflexive polytope ; 3 . @xmath167 is a gorenstein polytope of index @xmath0 in this case , @xmath177 @xmath178resp . @xmath176{}^{{\scriptstyle\vee}}})_{(r)}$]@xmath179 is a reflexive polytope with the unique interior lattice point @xmath173 ( resp . @xmath154 ) . [ reflcone ] the duality of reflexive gorenstein cones of index @xmath0 defines a natural duality of gorenstein polytopes of index @xmath0 : let @xmath109 be a @xmath1-dimensional gorenstein polytope of index @xmath0 . then the _ dual gorenstein polytope _ @xmath118 is defined as the support of the dual cone @xmath172{}^{{\scriptstyle\vee}}}$ ] , where @xmath150 is the reflexive gorenstein cone associated to @xmath101 . by [ reflcone ] , @xmath118 is also a @xmath1-dimensional gorenstein polytope of index @xmath0 . we want to compare the duality of @xmath1-dimensional gorenstein polytopes of index @xmath0 with the polar duality of reflexive polytopes . if @xmath180 , then the dualities are the same . for any @xmath181 two gorenstein polytopes @xmath101 and @xmath118 are combinatorially dual to each other . however , if @xmath182 , then the two reflexive polytopes @xmath183 and @xmath184 are _ not _ dual to each other as reflexive polytopes . in order to obtain the dual reflexive polytope @xmath185 from the gorenstein polytope @xmath118 one must replace the lattice @xmath186 by another one ( see [ refine ] and [ refldual ] below ) . let us consider the duality of gorenstein polytopes of index @xmath0 in more detail . let @xmath158 be a gorenstein cone of index @xmath0 with support @xmath187 . we define two lattices @xmath188 the lattice @xmath189 is dual to @xmath190 and one has @xmath191 and @xmath192 . [ refine ] the cone @xmath150 considered with respect to the new lattice @xmath189 is a gorenstein cone of index @xmath121 , and the reflexive polytope @xmath193 is its support . thus , we obtain : let @xmath187 be a gorenstein polytope of index @xmath0 and @xmath194 be the dual gorenstein polytope . if @xmath158 is a reflexive gorenstein cone of index @xmath0 with support @xmath187 , then the dual reflexive polytope @xmath195 equals @xmath196 with respect to the refined affine lattice @xmath197 [ refldual ] let @xmath198 be a facet of @xmath172{}^{{\scriptstyle\vee}}}$ ] . we denote by @xmath199 the minimal linear subspace in @xmath200 containing @xmath198 . then @xmath201 in particular , one has @xmath202{}^{{\scriptstyle\vee}}}\cap \nq = \rand \sigma^{\makebox[0mm]{}^{{\scriptstyle\vee}}}\cap \nqs\ ] ] and @xmath203 [ boundarylemma ] let @xmath204 . since @xmath198 is a facet of @xmath172{}^{{\scriptstyle\vee}}}$ ] , there is a vertex @xmath205 of @xmath167 such that @xmath206 for all @xmath207 . from @xmath208 we get @xmath209 . hence @xmath210 . now , @xmath211 implies @xmath212 . from the combinatorial point of view , the duality of gorenstein polytopes `` coincides '' with the polar duality . however , there exists a `` subtlety '' coming from the occuring lattices . let @xmath167 be a gorenstein polytope of index @xmath213 . then the gorenstein polytope @xmath196 has no interior lattice points . proposition [ boundarylemma ] shows that the reflexive polytope @xmath195 and @xmath196 have the same sets of boundary lattice points . assume that @xmath0 is not a prime number . let @xmath214 be any proper divisor of @xmath0 . then @xmath215 is a gorenstein polytope of index @xmath216 . the dual gorenstein polytope @xmath217 has the same lattice points on the boundary as @xmath195 and @xmath196 . therefore , two gorenstein polytopes @xmath217 and @xmath196 have the same combinatorial structure and also the same lattice point structure . however , @xmath217 and @xmath196 are gorenstein polytopes of different index , so they are _ not _ isomorphic as lattice polytopes . consider an example of this phenomenon : let @xmath101 be a unimodular @xmath78-dimensional simplex , i.e. , it has lattice volume @xmath121 . then @xmath101 is a gorenstein polytope of index @xmath218 . we have @xmath219 , i.e. @xmath101 is a selfdual gorenstein polytope . the dual reflexive polytope @xmath220 is a @xmath78-dimensional reflexive simplex with @xmath218 lattice points on the boundary . now consider @xmath221 as a gorenstein polytope of index @xmath134 . the dual gorenstein polytope @xmath222 is a gorenstein simplex of index @xmath134 having lattice volume @xmath134 and having only @xmath218 vertices as lattice points ( the same number of lattice points as in @xmath118 ) . the polytope @xmath223 is a reflexive simplex with @xmath224 lattice points on the boundary . the polytope @xmath225 is a self - dual reflexive simplex with @xmath226 lattice points on the boundary . pictures of these polytopes are given in example [ fig1 ] where the polytope @xmath227 is denoted by @xmath196 . in this section we consider cayley polytopes of length @xmath0 and give a criterion for a gorenstein polytope @xmath167 of index @xmath0 to be a cayley polytope of length @xmath0 . for this purpose we use the notion of a special @xmath5-dimensional simplex in the dual gorenstein polytope @xmath196 . let @xmath228 be dual lattices of rank @xmath1 ; and let @xmath229 be lattice polytopes . consider the lattice @xmath230 , where @xmath59 is the standard lattice basis of @xmath231 . then @xmath144 contains the affine sublattice @xmath232 . we identify @xmath233 with @xmath234 for @xmath235 . the polytope @xmath236 is called the _ cayley polytope of length @xmath0 _ associated to @xmath61 . the cone @xmath237 is called the associated _ cayley cone_. [ standard ] if @xmath238 , then the associated cayley cone is a gorenstein cone of dimension @xmath239 , and @xmath240 is its support . let @xmath241 be the standard lattice basis . then @xmath242 is called an _ unimodular simplex _ of dimension @xmath243 . let @xmath158 be a gorenstein cone with support @xmath187 . then the following statements are equivalent : 1 . @xmath150 is a cayley cone associated to @xmath0 lattice polytopes ; 2 . @xmath167 is a cayley polytope of length @xmath0 ; 3 . there is a lattice projection @xmath244 , which maps @xmath167 surjectively on @xmath245 ; 4 . there are nonzero @xmath246{}^{{\scriptstyle\vee}}}\cap \nq$ ] , such that @xmath247 moreover , the lattice vectors @xmath248 in @xmath249 form a part of a basis of @xmath145 and the cayley structure of @xmath167 is uniquely determined by @xmath0 polytopes @xmath250 these polytopes have properties @xmath251 , and @xmath252 . [ char ] ( 1 ) , ( 2 ) , ( 3 ) are obviously equivalent . \(1 ) @xmath253 ( 4 ) : let @xmath254 be a basis of the lattice @xmath144 , where @xmath255 is a basis of the lattice @xmath256 and @xmath59 the standard basis of @xmath231 . consider the dual basis @xmath257 of the lattice @xmath145 . then we have @xmath258 for @xmath259 and @xmath260 ( @xmath261 denotes the kronecker symbol , i.e. , @xmath262 , for @xmath263 , and @xmath116 otherwise ) . therefore we obtain @xmath246{}^{{\scriptstyle\vee}}}$ ] , @xmath264 . the latter implies @xmath265 . \(4 ) @xmath253 ( 1 ) : define the polytopes @xmath266 for @xmath267@xmath268 for @xmath269 . let @xmath270 be a vertex of @xmath167 . then @xmath271 . since @xmath272 for @xmath235 , we get that @xmath273 for some @xmath274 and @xmath275 . hence @xmath276 . take any @xmath54 . we claim that @xmath277 . indeed , otherwise we would have @xmath278 . this would contradict the condition @xmath279 . hence the mapping @xmath280 gives a projection of lattices @xmath244 such that the image of @xmath167 is @xmath245 . in particular , @xmath248 form a part of a basis of @xmath145 . recall the following definition from ( * ? ? ? 3.9 ) : a reflexive gorenstein cone @xmath150 of index @xmath0 is said to be _ completely split _ , if @xmath150 is the cayley cone associated to @xmath0 lattice polytopes . by [ char ] , we obtain the following characterization of completely split reflexive gorenstein cones : let @xmath158 be a reflexive gorenstein cone of index @xmath0 with support @xmath167 . then the following statements are equivalent : 1 . @xmath150 is completely split ; 2 . there exist lattice points @xmath281 such that @xmath282 moreover , there is a @xmath121-to-@xmath121 correspondence between all possible collections @xmath283 of @xmath0 lattice points in @xmath196 such that @xmath284 and all possible cayley polytope structures of length @xmath0 of the gorenstein polytope @xmath167 . [ compl ] \(1 ) @xmath253 ( 2 ) : assume that @xmath150 is completely split . by [ char](4 ) there exist lattice points @xmath246{}^{{\scriptstyle\vee}}}$ ] such that @xmath265 . since @xmath285 and @xmath286 for all @xmath235 , we get @xmath287 for all @xmath235 . thus , all lattice points @xmath248 belong to @xmath196 . \(2 ) @xmath253 ( 1 ) : the statement follows immediately from [ char ] , because the index of @xmath150 equals @xmath288 . by [ char ] , every completely split reflexive gorenstein cone @xmath147 determines @xmath0 lattice polytopes @xmath289 . on the other hand , one can start with @xmath0 lattice polytopes @xmath290 and ask whether the corresponding cayley cone is reflexive of index @xmath0 , or whether the cayley polytope @xmath291 is a gorenstein polytope of index @xmath0 . the answer is contained in ( * ? ? ? * prop.3.6 ) : let @xmath292 be lattice polytopes such that the minkowski sum @xmath293 has dimension @xmath1 . then the following statements are equivalent : 1 . the associated cayley cone @xmath150 is reflexive of index @xmath0 ; 2 . @xmath291 is a gorenstein polytope of index @xmath0 ; 3 . @xmath293 is a reflexive polytope ( with interior lattice point @xmath53 ) . in this case , the lattice point @xmath294 ( see def . [ refl - g ] ) equals @xmath295 . [ sum ] for convenience , we give a purely combinatorial proof of this statement . \(1 ) @xmath296 ( 2 ) : follows from prop . [ reflcone ] . \(2 ) @xmath253 ( 3 ) : let @xmath297 be a gorenstein polytope of index @xmath0 . denote by @xmath298 the projections @xmath299 and @xmath300 . since @xmath301 is reflexive , there exists a unique lattice point @xmath302 . moreover , the @xmath298-image of @xmath53 can only be the unique interior point @xmath303 in the reflexive simplex @xmath304 ( where @xmath305 ) . we remark that the intersection of the affine subspace @xmath306 with @xmath301 is exactly @xmath307 . since @xmath53 has integral distance @xmath121 from any facet of @xmath301 , the lattice point @xmath53 has also integral distance @xmath121 from any facet of @xmath307 ( facets of @xmath307 are intersections of facets of @xmath301 with the affine subspace @xmath306 ) . \(3 ) @xmath253 ( 1 ) : assume that the lattice polytope @xmath308 is reflexive . denote by @xmath53 the unique interior lattice point in @xmath3 . consider the cayley cone @xmath150 over @xmath297 and denote by @xmath173 the lattice point @xmath295 in @xmath309 . we have to show that @xmath310 for any primitive lattice vector @xmath311 generating a @xmath121-dimensional face @xmath312 of the dual cone @xmath172{}^{{\scriptstyle\vee}}}\subset \nrq$ ] . let @xmath313 the facet of @xmath150 which is orthogonal to @xmath314 . there are the following two possibilities : case 1 . there exists @xmath315 such that @xmath316 . then @xmath317 is orthogonal to @xmath318 , because @xmath317 is generated by some lattice points from the polytopes @xmath319 @xmath320 and @xmath321 for @xmath322 . since @xmath318 is primitive and @xmath323{}^{{\scriptstyle\vee}}}$ ] , we get @xmath324 and @xmath325 . case 2 . for any @xmath315 there exists a lattice point @xmath326 which is contained in @xmath317 . in this case , @xmath317 is a cayley cone of @xmath0 lattice polytopes @xmath327 @xmath328 such that @xmath329 @xmath330 . moreover , @xmath331 is a @xmath22-dimensional face of @xmath332 . let @xmath333 be the @xmath334-dimensional sublattice in @xmath144 defined by @xmath243 equations @xmath335 . then @xmath336 is exactly the cone over the @xmath1-dimensional reflexive polytope @xmath337 . moreover , @xmath173 is the corresponding interior lattice point in @xmath338 . there exists a primitive lattice vector @xmath339 in the dual lattice @xmath340 such that @xmath341{}^{{\scriptstyle\vee}}}$ ] , @xmath342 , and @xmath343 . we identify the dual lattice @xmath340 with @xmath344 . since @xmath314 and @xmath339 have the same orthogonal subspace in @xmath345 there exists a positive integer @xmath156 such that @xmath346 modulo the sublattice generated by @xmath347 . in this case , we have @xmath348 , because @xmath349 for all @xmath350 . now it remains to show that @xmath351 . using the splittings @xmath352 we can write @xmath353 for some integers @xmath354 . applying @xmath355 to both sides of the last equation , we get @xmath356 i.e. , @xmath156 divides @xmath357 for all @xmath358 . since @xmath359 is a primitive lattice vector , we get @xmath351 . our next purpose is to explain the relations between cayley cones and cayley polytopes to some results in @xcite . it would be more convenient to work with polytopes in the dual space @xmath360 . for this reason , we denote these polytopes by @xmath361 . athanasiadis introduced in @xcite the concept of a special simplex @xmath362 of a convex polytope @xmath361 . here we slightly modify this definition by demanding the vertices of the simplex @xmath362 to be lattice points in @xmath361 . let @xmath363 be a @xmath1-dimensional lattice polytope . a simplex @xmath362 spanned by @xmath0 affinely independent lattice points in @xmath361 is called a _ special @xmath5-simplex _ of @xmath361 , if each facet of @xmath361 contains exactly @xmath243 vertices of @xmath362 . now , we can give an addendum to cor . [ compl ] : let @xmath363 be a gorenstein polytope of index @xmath0 . there is a @xmath121-to-@xmath121 correspondence between cayley polytope structures of length @xmath0 of the dual gorenstein polytope @xmath364 and special @xmath5-simplices @xmath365 . a special @xmath5-simplex @xmath366 can be characterized as a lattice @xmath5-dimensional simplex in @xmath361 , which is not contained in the boundary of @xmath361 . moreover , all special @xmath5-simplices in @xmath361 have the same barycenter . [ ath ] let @xmath367 be the reflexive gorenstein cone associated to @xmath368 . we identify @xmath101 with the support @xmath167 of @xmath150 and @xmath361 with the support @xmath369 of @xmath172{}^{{\scriptstyle\vee}}}$ ] . if @xmath101 is a cayley polytope of length @xmath0 , then , by cor . [ compl ] , there exist @xmath370 such that @xmath371 . if @xmath124 is a facet of @xmath369 , then there is a vertex @xmath372 such that @xmath373 . let @xmath252 as in prop . then @xmath374 for some @xmath274 . this implies @xmath375 and @xmath376 for @xmath267 . so @xmath377 , but @xmath378 for @xmath267 . so @xmath379 is a special @xmath5-simplex in @xmath361 . now assume that @xmath370 are vertices of a @xmath5-dimensional simplex that is not contained in the boundary of @xmath361 . then @xmath380 is in the interior of @xmath172{}^{{\scriptstyle\vee}}}$ ] . since @xmath381 is reflexive , we have @xmath382 . therefore @xmath371 . by prop . [ char ] , @xmath101 is a cayley polytope of length @xmath0 . moreover , above arguments show that @xmath383 is a special @xmath5-simplex in @xmath361 . if @xmath384 is another special simplex in @xmath361 then @xmath385 thus @xmath362 and @xmath386 have the same barycenter . if @xmath363 is a reflexive polytope ( gorenstein polytope of index @xmath180 ) then there exists only one special @xmath5-simplex in @xmath361 which is exactly the unique interior lattice point in @xmath361 . [ r=1 ] let @xmath387 be the standard @xmath1-dimensional unit cube in @xmath388 : @xmath389 . then @xmath387 is a gorenstein polytope of index @xmath390 . there exist exactly @xmath391 different special @xmath121-simplices in @xmath387 which are spanned by all possible pairs of opposite vertices of @xmath387 . the dual gorenstein polytope @xmath392 has exactly @xmath1 special @xmath121-simplices which @xmath121-to-@xmath121 correspond to @xmath1 different cayley structures of the @xmath1-dimensional unit cube @xmath387 . [ cube ] let us recall the following definition : a lattice polytope @xmath393 is called _ integrally closed _ , if any lattice point in the gorenstein cone @xmath150 over @xmath361 is a sum of lattice points from the support @xmath394 . we get the following corollary : let @xmath393 be an integrally closed gorenstein polytope of index @xmath0 . then @xmath361 contains a special @xmath5-simplex and @xmath395 is a cayley polytope of length @xmath0 . [ normal ] it follows immediately from the fact that the lattice point @xmath154 is a sum of @xmath0 lattice points @xmath396 from @xmath369 . one can easily show that if a lattice polytope @xmath361 admits a unimodular triangulation , then @xmath361 is integrally closed . under the assumption that @xmath361 admits a unimodular triangulation the statement in [ normal ] was proved in ( * ? ? ? one of the main results in @xcite is corollary 7 , where bruns and roemer show that for any integrally closed gorenstein polytope there exists a reflexive polytope with the same @xmath397-vector . here we give a slightly more general version of this result that does not need any commutative algebra . we first recall the definition of the @xmath397-polynomial ( also called @xmath123-polynomial ) , cf . @xcite : let @xmath363 be a @xmath1-dimensional lattice polytope . @xmath398 of @xmath361 is defined as the numerator of the ehrhart series , i.e. , @xmath399 . [ h - po ] the @xmath397-polynomial has degree at most @xmath1 . it is well - known , see @xcite , that @xmath361 is gorenstein of index @xmath0 if and only if for the @xmath397-polynomial one has : @xmath400 in this case , @xmath361 is reflexive if and only if the degree of @xmath401 is @xmath1 . [ symm ] the following theorem is very close to results in @xcite , @xcite , and @xcite : let @xmath393 be a gorenstein polytope of index @xmath0 . assume that @xmath361 contains a special @xmath5-simplex with vertices @xmath396 , i.e. the dual gorenstein polytope @xmath395 is a cayley polytope associated with some lattice polytopes @xmath402 . then the projection of @xmath361 along @xmath403 is a @xmath404-dimensional reflexive polytope @xmath3 such that @xmath405 and @xmath406 [ brtheo ] the statement follows straightforward from ( * ? ? ? * lemma 4 , lemma 5 ) . another variant of the same ideas is contained in ( * ? ? ? * prop.3.12 ) and ( * ? ? ? * lemma 3.4 , cor . 4.1 , cor . 4.2 ) . for convenience , we explain the main combinatorial idea of the proof . let @xmath369 be the support of the gorenstein cone over @xmath361 . we identify @xmath361 with @xmath369 . for any @xmath407 let @xmath408 be the set of facets of @xmath369 that do _ not _ contain @xmath409 . we denote by @xmath317 the union of all faces @xmath410 , where @xmath411 for @xmath235 . take one such a face @xmath412 . we define a lattice polytope @xmath413 . let @xmath414 be the vertices corresponding to the facets @xmath415 . by prop . [ char ] , we have @xmath416 for @xmath259 . therefore @xmath417 is an @xmath0-fold pyramid over @xmath418 , i.e. , @xmath419 , cf . * remark 2.6 ) . by ( * lemma 4 ) the set of all @xmath417 ( together with their subfaces ) form a polyhedral decomposition of @xmath369 . denote by @xmath298 the projection of @xmath361 along @xmath403 . let @xmath420 be the unique interior point with lattice distance @xmath143 from any facet of @xmath361 . we define the refined lattice @xmath421 with @xmath422 $ ] by @xmath423 . by prop . [ ath ] and prop . [ char ] , we have @xmath424 . so the projection @xmath298 can be seen as the surjective map @xmath425 where @xmath426 form a part of @xmath427-basis of @xmath340 . let @xmath124 be a face of @xmath369 which is the preimage of a facet of @xmath3 . by definition of special @xmath5-simplex , any facet of @xmath369 is contained in @xmath428 . since @xmath377 for all @xmath235 @xmath429 , there exists @xmath411 which contains @xmath124 . thus @xmath430 for some @xmath431 . on the other hand , @xmath432 and @xmath433 . therefore , @xmath434 . by ( * lemma 5 ) , the projection @xmath435 respects two polyhedra decompositions @xmath436 @xmath437 where each @xmath438 is a @xmath0-fold pyramid over a face @xmath431 and the polytope @xmath439 is @xmath121-fold pyramid with vertex @xmath116 over a facet in @xmath440 . since the @xmath397-polynomials of @xmath438 and @xmath439 are the same , a standard inclusion - exclusion argument shows that the @xmath397-polynomials of @xmath369 and @xmath3 are the same . using [ symm ] , this implies that @xmath3 is a reflexive polytope . by theorem [ sum ] , @xmath293 is a reflexive polytope . moreover , the proof of theorem [ sum ] shows that @xmath3 is the dual reflexive polytope to @xmath293 . the above theorem is illustrated in the following example : consider one - dimensional lattice polytopes @xmath441 and @xmath442 . we remark that @xmath443 contains @xmath116 , but @xmath444 does not . the cayley polytope @xmath445 is a gorenstein polytope of index @xmath134 . on the other hand , @xmath196 is isomorphic to @xmath446 , where @xmath447 is the unimodular @xmath78-simplex . the polytope @xmath196 contains a special @xmath121-simplex @xmath362 corresponding to the cayley polytope structure of @xmath167 ( the two vertices of @xmath362 are marked by boxes ) . by projecting along this simplex , we obtain the reflexive square @xmath3 . the boundary subcomplex @xmath317 of @xmath196 ( which maps bijectively to the boundary of @xmath3 ) is marked by thick edges . the dual reflexive polytope @xmath4 of @xmath3 is @xmath448 . it is the intersection of @xmath449 with an affine plane which is orthogonal to the special simplex @xmath362 . + [ fig1 ] now we are interested in the case when both dual reflexive gorenstein cones are completely split . using special simplices , we want to obtain a simple combinatorial characterization of nef - partitions and their duality . let us consider the following purely combinatorial definition of a nef - partition ( cf . * prop.3.2 ) ) : let @xmath450 be a @xmath1-dimensional reflexive polytope and @xmath451 . a minkowski sum decomposition @xmath337 where @xmath452 are lattice polytopes is called a _ nef - partition _ of @xmath101 of length @xmath0 , if there exist lattice points @xmath453 such that @xmath454 . since the reflexive polytope @xmath101 is determined by @xmath61 , for simplicity , we the name _ nef - partition _ just for the the set of lattice polytopes @xmath61 . we call a nef - partition _ centered _ , if @xmath455 for all @xmath456 . we call a nef - partition _ proper _ , if @xmath457 for all @xmath456 . [ nefdef ] let @xmath13 be the @xmath1-dimensional reflexive cube defined by the conditions @xmath458 @xmath459 . then @xmath101 is a minkowski sum of two unit cubes @xmath444 and @xmath443 defined as @xmath460 @xmath461 since @xmath462 @xmath463 , the minkowski sum @xmath464 defines a proper centered nef - partition of length @xmath134 . this example comes from the gorenstein polytopes considered in [ cube ] . [ nefp ] it clear that if @xmath337 is a nef - partition of @xmath101 , then @xmath465 is a centered nef - partition . moreover , if @xmath466 is a centered nef - partition of @xmath101 , then @xmath467 is a proper centered nef - partition . therefore , the study of arbitrary nef - partitions immediately reduces to study of proper centered nef - partitions . the above definition of proper centered nef - partitions is nearly precisely the one given by kreuzer , riegler and sahakyan in ( * ? ? ? * prop.3.2 ) . they also demand an additional assumption @xmath468 for @xmath469 , which is actually unnecessary ( see cor . [ inter ] ) . from the viewpoint of algebraic geometry , proper centered nef - partitions were introduced by borisov in @xcite . let @xmath470 be a gorenstein toric fano variety corresponding to the reflexive polytope @xmath101 which contains @xmath116 in its interior . denote by @xmath471 the @xmath1-dimensional torus acting on @xmath470 ( we identify @xmath471 with its open dense orbit in @xmath470 ) . a minkowski sum decomposition @xmath337 describes the anticanonical sheaf @xmath472 as a tensor product of @xmath0 @xmath471-equivariant semi - ample line bundles @xmath473 . in this case , the lattice points @xmath474 can be identified with some global @xmath471-invariant sections @xmath475 of @xmath476 which do not vanish on @xmath477 . the condition @xmath478 means that if @xmath479 is the zero set of the global section @xmath475 @xmath480 , then @xmath481 is the anticanonical divisor of @xmath470 which contains every irreducible component of @xmath482 with multiplicity @xmath121 . in particular , @xmath479 and @xmath483 do not have common irreducible components for @xmath484 . moreover , @xmath485 if and only if @xmath486 . therefore , if @xmath485 for all @xmath456 , then we get a _ partition _ of the set of irreducible components of @xmath482 into @xmath0 nonempty subsets which define _ nef - divisors _ @xmath487 such that @xmath481 is the anticanonical divisor . this explains the origin of the notion _ nef - partition _ in @xcite . we characterize nef - partitions in the following way : let @xmath101 be a reflexive polytope with the unique interior lattice point @xmath53 , and let @xmath61 be lattice polytopes such that @xmath337 . denote by @xmath150 the cayley cone associated to @xmath61 . then the following statements are equivalent : 1 . the dual cone @xmath172{}^{{\scriptstyle\vee}}}$ ] is a completely split reflexive gorenstein cone of index @xmath0 2 . @xmath291 is a gorenstein polytope of index @xmath0 containing a special @xmath5-simplex 3 . there exist lattice points @xmath474 @xmath488 such that @xmath454 , i.e. , @xmath489 is a nef - partition . [ splittheo ] \(1 ) @xmath253 ( 2 ) : if @xmath172{}^{{\scriptstyle\vee}}}$ ] is a reflexive gorenstein cone of index @xmath0 , then this holds also for @xmath150 . hence theorem [ sum ] implies that @xmath291 is a gorenstein polytope of index @xmath0 . since @xmath172{}^{{\scriptstyle\vee}}}$ ] is completely split , prop . [ ath ] implies that @xmath291 contains a special @xmath5-simplex . \(2 ) @xmath253 ( 3 ) : since @xmath172{}^{{\scriptstyle\vee}}}$ ] is completely split , cor . [ compl ] implies that there are @xmath490 such that @xmath491 . let @xmath492 as in prop . [ char ] and cor . [ compl ] . since @xmath493 @xmath494 and @xmath495 is a nonnegative integer @xmath496 , every polytope @xmath497 contains exactly one point from @xmath498 . without loss of generality we can assume that @xmath499 @xmath494 . we can write @xmath500 @xmath501 . by [ sum ] , @xmath502 . hence for @xmath269 we have @xmath503 with @xmath454 and @xmath61 is a nef - partition . \(3 ) @xmath253 ( 1 ) : by theorem [ sum ] we know that @xmath150 is a reflexive gorenstein cone of index @xmath0 , hence also @xmath172{}^{{\scriptstyle\vee}}}$ ] is . we define @xmath504 for @xmath269 . then @xmath505 with @xmath491 . hence @xmath172{}^{{\scriptstyle\vee}}}$ ] is completely split by cor . [ compl ] . we can sum this up in a more symmetric way : a reflexive gorenstein cone @xmath150 is associated to a nef - partition if and only if both cones @xmath150 and @xmath172{}^{{\scriptstyle\vee}}}$ ] are completely split . a gorenstein polytope @xmath167 of index @xmath0 is the cayley polytope of a nef - partition if and only if @xmath167 and @xmath196 are cayley polytopes of length @xmath0 , or equivalently , if and only if both polytopes @xmath167 and @xmath196 contain special @xmath5-simplices . [ splitcoro ] in example [ fig1 ] , the polytope @xmath506 is a gorenstein polytope of index @xmath134 . it does not contain a special @xmath121-simplex . thus @xmath196 is not a cayley polytope of length @xmath134 . in other words , the cayley cone @xmath150 associated to @xmath507 is completely split , while @xmath172{}^{{\scriptstyle\vee}}}$ ] is not . hence the sum @xmath448 is not a nef - partition of the reflexive polytope @xmath448 , because there are no lattice points @xmath508 , @xmath509 such that @xmath510 . from cor . [ normal ] we get : let @xmath167 be a gorenstein polytope such that both @xmath167 and @xmath196 are integrally closed . then @xmath167 ( and also @xmath196 ) is a cayley polytope associated to a nef - partition . in dimension @xmath134 , all lattice polygons @xmath507 such that @xmath448 is reflexive were listed in ( * ? ? ? * fig.4,5 ) . in dimension @xmath78 and @xmath218 , all reflexive polytopes are completely classified in @xcite . using the software package palp , one can find all possible centered nef - partitions whose minkowski sum yields a given reflexive polytope @xcite . however , it would be desirable to have an efficient algorithm that classifies all centered nef - partitions of given length @xmath0 in fixed dimension @xmath1 . at the moment , the only way is to compute _ all _ reflexive polytopes of dimension @xmath1 , which is practically impossible for @xmath511 . let @xmath3 and @xmath4 be two dual to each other @xmath1-dimensional gorenstein polytopes of index @xmath0 . assume that there exist two special @xmath512-simplices @xmath6 and @xmath7 . by [ ath ] , these simplices define cayley structures on polytopes @xmath3 and @xmath4 . therefore , for some lattice polytopes @xmath61 and @xmath513 we obtain @xmath514 and two nef - partitions @xmath515 of @xmath404-dimensional reflexive polytopes @xmath101 and @xmath361 . we call these two nef - partitions _ dual to each other_. [ nef - part - dual ] the duality of nef - partitions defined in [ nef - part - dual ] looks trivial , because it simply exchanges gorenstein polytopes @xmath3 with @xmath4 and special @xmath5-simplices @xmath362 and @xmath386 . by [ r=1 ] , for @xmath2 this duality reduces to the polar duality between reflexive polytopes @xmath3 and @xmath4 . however , definition [ nef - part - dual ] is not very convenient if we want to determine explicitly the polytopes @xmath516 from polytopes @xmath61 . for this reason , we consider below another combinatorial approach to the duality of centered nef - partitions . let @xmath99 be a reflexive polytope and @xmath517 . assume that @xmath518 is a centered nef - partition . then @xmath0 polytopes @xmath519 define the dual centered nef - partition . [ nefdualdef ] consider two lattices @xmath520 , and @xmath521 , where @xmath522 . let @xmath158 be the reflexive cayley cone associated with the lattice polytopes @xmath229 . denote by @xmath172{}^{{\scriptstyle\vee}}}\subset \nrq$ ] the dual reflexive cone . we set @xmath497 . then @xmath523 , and @xmath524 . let @xmath196 be the support of @xmath172{}^{{\scriptstyle\vee}}}$ ] . for every @xmath235 , we define polytopes @xmath525 for @xmath267@xmath268 . by prop . [ char ] , we have @xmath526 . by prop . [ splittheo ] , for every @xmath235 , we can write @xmath527 as @xmath528 for some @xmath529 . we want to show that @xmath530 @xmath501 . this follows from the observation that for @xmath531 and @xmath54 one has @xmath532 if and only if @xmath533 ( i.e. , @xmath534 @xmath535 @xmath536 ) . obviously , one has @xmath537 @xmath488 . so we obtain that @xmath538 is the centered dual nef - partition . [ dual - nefp ] the dual centered nef - partition to the one considered in [ nefp ] consists of two @xmath1-dimensional standard simplices @xmath539 @xmath540 the following picture below describes another example of two dual to each other nef - partitions in dimension @xmath134 . + in this case , the dimensions of @xmath443 and @xmath541 are not maximal . [ fig3 ] [ nef - p2 ] for the remainder of this section let @xmath542 and @xmath513 be two dual to each other centered nef - partitions . for any @xmath543 one has @xmath544 if and only if @xmath545 . in particular , the dual nef - partition of a proper nef - partition is a proper nef - partition . [ proper ] without loss of generality , we can assume that @xmath546 . then the minkowski sum @xmath547 is a reflexive polytope @xmath101 containing @xmath116 in its interior . hence @xmath548 , so by the formulas in [ nefdualdef ] we obtain that @xmath549 . by duality of nef - partitions , the condition @xmath549 implies @xmath546 , analogously . with the same argument we also derive from prop . [ nefdualdef ] : @xmath550 for @xmath469 . [ inter ] the crucial property of the duality of nef - partitions is that it exchanges minkowski sum and convex hull : @xmath551 is a reflexive polytope which is dual to @xmath552 . moreover , the two lattice polytopes @xmath291 and @xmath551 have the same @xmath397-polynomial . [ minkdual ] the statements follow immediately from theorem [ brtheo ] , because @xmath551 is the projection of @xmath291 along @xmath553 . hence , we have for @xmath337 and @xmath554 : @xmath555 now , we show that in order to determine the dual nef - partition it suffices to dualize the minkowski sum @xmath101 , and to choose the vertices of @xmath513 among the vertices of @xmath118 . for this we take a closer look at boundary of the lattice polytopes @xmath513 : we have for @xmath235 : @xmath556 any face @xmath124 of @xmath557 , with @xmath558 , is a face of @xmath559 and contained in a face of @xmath554 . [ faceprop ] let @xmath560 . then @xmath561 and these are all numbers in @xmath562 $ ] . from prop . [ nefdualdef ] the equation follows . now , let @xmath124 be a facet of @xmath557 , with @xmath558 . let @xmath205 be a point in the relative interior of @xmath124 . by prop . [ nefdualdef ] we have @xmath563 , so there is a vertex @xmath564 of @xmath565 such that @xmath566 . therefore @xmath567 , and @xmath124 is also contained in the face @xmath568 of @xmath552 . we show that @xmath569 . let @xmath570 with @xmath571 . there is a convex combination @xmath572 , with @xmath573 , @xmath574 , for @xmath575 ( @xmath576 ) . hence from @xmath577 we get @xmath578 , because of prop . [ nefdualdef ] , thus @xmath579 , so @xmath580 . this shows @xmath581 . the reader further interested in faces of nef - partitions and their dualities is encouraged to consult ( * ? ? ? * prop.2.3 ) and @xcite . as mentioned above , here is a direct way to determine the vertices of the dual nef - partition : we have for @xmath235 : @xmath582 [ nabcor ] the previous statement implies that any non - zero vertex of @xmath557 ( @xmath235 ) is a vertex of @xmath583 . it is not difficult to prove , that this also holds under the weaker assumption that @xmath513 are lattice polytopes , each containing @xmath116 , such that @xmath552 has no interior lattice point except @xmath116 . not only vertices but also lattice points of @xmath584 necessarily belong to one of the lattice polytopes @xmath513 : any non - zero lattice point of @xmath118 is contained for some @xmath54 in a face of @xmath557 that does not contain the origin . in particular @xmath585 . let @xmath586 , @xmath587 . there has to be some @xmath588 ( @xmath54 ) with @xmath589 , thus @xmath566 . now use prop . [ faceprop ] . borisov and the first author gave in @xcite an explicit formula for the generating function of stringy hodge numbers of calabi - yau complete intersections given by a nef - partition ( the stringy hodge numbers were introduced in @xcite ) . we remark that this formula can be used for arbitrary reflexive gorenstein cones or for arbitrary gorenstein polytopes . however , it is not clear a priori whether the corresponding @xmath590-function is a polynomial . we expect that it is always the case . in this section , we review the above formula using @xmath591-polynomials and ideas from @xcite . our goals are to introduce the @xmath590-function of a gorenstein polytope , and to show some of its properties , and to formulate related open questions and conjectures . we start by recalling the definition of the @xmath592- and @xmath593-polynomials , defined by stanley for general eulerian posets in @xcite ( see also ( * ? ? ? * def.5.3 ) ) : let @xmath594 be an eulerian poset of rank @xmath1 , @xmath595 its minimal element , @xmath596 its maximal element . then the polynomials @xmath597 $ ] are defined recursively as follows . if @xmath598 then we set @xmath599 . if @xmath600 then the @xmath593-polynomial is defined as @xmath601}(t).\ ] ] for @xmath600 , we define the @xmath592-polynomial @xmath602 as a polynomial of degree @xmath603 whose coefficients @xmath604 ( for @xmath605 ) are equal to the coefficients @xmath606 of the polynomial @xmath607 i.e. , @xmath608 for @xmath605 . a recent account on this important combinatorial notion can be found in @xcite . here we need only the following well - known properties : the polynomials @xmath609 and @xmath610 have the followng properties : 1 . every coefficient of @xmath609 is a non - negative integer ; 2 . @xmath611 if and only if the poset @xmath594 is boolean , i.e. , it is isomorphic to the poset of all faces ( including the empty one ) of a @xmath22-dimensional simplex ; 3 . the polynomial @xmath592 is multiplicative , i.e. , @xmath612 . combinatorial data contained in the @xmath592-polynomial will be combined with the enumeration of lattice points via the @xmath397-polynomial in the following definition ( * ? ? ? * def.5.3 ) : let @xmath3 be a lattice polytope . we define the @xmath591-polynomial : @xmath613}(t),\ ] ] where @xmath614 is the @xmath397-polynomial of the lattice polytope @xmath124 ( with convention @xmath615 , see def . [ h - po ] ) and @xmath616}(t)$ ] is the @xmath592-polynomial of the poset @xmath617 $ ] of faces between @xmath124 and @xmath3 . note that @xmath618 . however , @xmath619 , if @xmath620 . the @xmath591-polynomial of a lattice polytope @xmath3 has the following properties : 1 . all coefficients of @xmath621 are non - negative integers ( this follows from ( * ? ? ? * prop.5.5 ) , where it is shown that @xmath621 is the hilbert function of a graded vector space ) . 2 . one has the reciprocity - law ( see ( * ? ? ? * remark 5.4 ) ) : @xmath622 here is another nice property of the @xmath591-polynomial : let @xmath3 be a lattice polytope . if @xmath3 is a lattice pyramid , then @xmath619 . [ pyra ] let @xmath3 be a lattice pyramid over the facet @xmath623 with apex @xmath205 , i.e. , @xmath624 . any face @xmath124 of @xmath3 , that is not contained in @xmath623 , is a pyramid over a face @xmath625 of @xmath623 with apex @xmath205 . hence , @xmath626 and @xmath616}(t ) = g_{[f',p']}(t)$ ] . on the other hand , if @xmath124 is a face of @xmath623 , then @xmath627}(t ) = g_{[f',p']}(t)$ ] , since @xmath628 $ ] is a product of @xmath629 $ ] and an interval , and @xmath592 is multiplicative . this yields : @xmath630 in the case of a simplex , the @xmath591-polynomial is easy to calculate : let @xmath631 be a lattice simplex . let @xmath632 . we denote by @xmath633 the parallelepiped spanned by the vertices of @xmath634 . then @xmath635 the degree of @xmath621 equals at most the degree of @xmath636 . [ simplexprop ] for a simplex @xmath634 , the @xmath397-polynomial can be calculated by counting the lattice points in the parallelepiped @xmath637 , where @xmath638 are the vertices of @xmath634 : @xmath639 since the @xmath592-polynomial of a simplex equals @xmath121 , we get @xmath640 now , for @xmath641 , we denote by @xmath642 the smallest face of @xmath634 such that @xmath643 . hence , @xmath644 the euler - poincar - formula implies that the right side vanishes , whenever @xmath645 . from this the equation follows . in general , the degree of the @xmath591-polynomial of a lattice simplex is not equal to the degree of the @xmath397-polynomial . this follows immediately from lemma [ pyra ] . for any lattice polytope @xmath646 we denote by @xmath647 ( resp . by @xmath648 ) the number of @xmath256-lattice points in the relative interior of @xmath3 ( resp . in @xmath649 ) . now , let @xmath3 be a @xmath134-dimensional lattice polytope with @xmath156 vertices ( a lattice @xmath156-gon ) . in this case , we calculate from the definition of the @xmath593- and @xmath592-polynomial , @xmath650}(t ) = k + k ( t-1 ) + ( t-1)^2 = 1 + ( k-2)t + t^2 $ ] , thus @xmath651}(t ) = 1 + ( k-3 ) t$ ] . then @xmath652 @xmath653 @xmath654 since @xmath655 more generally , if @xmath3 is a @xmath1-dimensional lattice polytope and @xmath656 , then @xmath657 and @xmath636 have the same leading term @xmath658 , because @xmath659}(t ) ) \leq d-1\ ] ] for any proper face @xmath660 , since @xmath661 and @xmath662}(t ) ) < ( \rank [ f , p])/2 = ( d-\dim(f))/2 $ ] . let @xmath101 be a @xmath1-dimensional gorenstein polytope of index @xmath0 . we define the following function of two independent variables @xmath564 and @xmath205 : @xmath663 where @xmath664 is the face of @xmath118 dual to @xmath124 . we call it the _ stringy @xmath665-function _ ( or @xmath590-function ) of @xmath101 . the integer @xmath666 we call the _ calabi - yau dimension _ ( cy - dimension ) of @xmath101 . 1 . the @xmath590-function is symmetric @xmath667 and it satisfies the poincar duality @xmath668 2 . the @xmath590-function satisfies the following reciprocity - law : @xmath669 note , that we have by definition @xmath670 . 3 . let @xmath671 be the degree of the @xmath397-polynomial of @xmath101 . then we have @xmath672 , cf . 4 . the cy - dimension can be arbitrarily negative : if @xmath101 is a lattice pyramid over @xmath673 , then @xmath674 , since the index of @xmath101 is @xmath121 plus the index of @xmath673 . [ erez ] here is our conjecture on the @xmath590-function : let @xmath101 be a gorenstein polytope of cy - dimension @xmath675 . then @xmath676 is a polynomial in @xmath677 of degree @xmath678 with nonnegative integral coefficients @xmath679 ( in particular , one has @xmath680 if @xmath681 , and @xmath665 is constant if @xmath682 ) . moreover , for any @xmath683 the polynomial @xmath665 satisfies the following two conditions @xmath684 and @xmath685 these two conditions together with ( 1 ) and ( 2 ) from [ erez ] imply : 1 . if @xmath686 , then for some integer @xmath156 one has @xmath687 2 . if @xmath688 , then for some integers @xmath689 one has @xmath690 in particular , the number @xmath691 is always divisible by @xmath692 . [ e - conj ] this conjecture is motivated by the expectation that the function @xmath693 `` looks '' as a generating function for stringy hodge numbers of a disjoint union of manifolds of dimension @xmath694 with trivial canonical class . the second condition was proved for stringy hodge numbers in a more general form in @xcite . in the case @xmath695 we expect a finite union of points . in the case @xmath696 , it should be a union of elliptic curves . in the case @xmath697 , one expects hodge numbers of abelian surfaces and @xmath698-surfaces . in all examples that the authors know the leading coefficient @xmath156 is either zero , or a power of @xmath134 ( see also [ alggeonefpart ] ) . let @xmath699 be a standard unimodular simplex of dimension @xmath700 @xmath701 ) . we calculate the @xmath590-function of the gorenstein simplex @xmath702 which has index @xmath0 . we have @xmath703 . by lemma [ boundarylemma ] the dual gorenstein polytope @xmath118 is a gorenstein simplex with only unimodular facets . hence , lemma [ pyra ] yields : @xmath704 we claim : @xmath705 combining these two equations we get : @xmath706 we show ( [ seq ] ) , by proving that @xmath707 for any gorenstein simplex @xmath101 of index @xmath0 and with @xmath703 . by prop . [ simplexprop ] and its proof , to determine @xmath708 we have to compute the sum @xmath709 over the lattice points @xmath270 in the interior of the parallelepiped spanned by vertices of @xmath101 . on the other hand , in order to determine @xmath710 one computes the sum @xmath709 over _ all _ lattice points @xmath270 in the paralellepiped . by remark [ erez](3 ) , the degree of @xmath711 equals the index @xmath0 . hence , there is only one interior lattice point @xmath270 in the parallelepiped such that @xmath712 . in this case , @xmath270 is the unique interior lattice point in @xmath713 which is given as @xmath714 . hence @xmath715 . in the same way , this holds for @xmath118 . this proves ( [ seq ] ) . from remark [ erez](4 ) we see that using lattice pyramids we may construct gorenstein polytopes of negative cy - dimension . in this case , the @xmath590-function should vanish . however , this vanishing holds for any lattice pyramid ( even of positive cy - dimension ) : let @xmath101 be a gorenstein polytope . if @xmath101 is a lattice pyramid , then @xmath716 . [ pyramid ] let @xmath101 be a lattice pyramid over a facet @xmath673 with apex @xmath205 . then also @xmath118 is a lattice pyramid over the facet @xmath717 with apex @xmath718 , for instance , use prop . thus , for any face @xmath124 of @xmath101 , @xmath124 or @xmath664 is a lattice pyramid . hence , @xmath719 by lemma [ pyra ] . from this the statement follows . originally , in @xcite so - called _ @xmath720-polynomials _ were used to calculate hodge numbers , see also ( * ? ? ? * remark 5.8 ) and @xcite . in ( * ? ? ? * lemma 11.4 ) borisov and mavlyutov gave the following definition in terms of @xmath592-polynomials : let @xmath594 be an eulerian poset of rank @xmath1 , with minimal element @xmath595 and maximal element @xmath596 . then @xmath721^}(u^{-1 } v)\ ; g_{[\hat{0},x]}(uv).\ ] ] to see , how the @xmath720-polynomial naturally appears , we separately collect all @xmath592-polynomials in the definition of the @xmath590-function into one polynomial in two variables : let @xmath101 be a gorenstein polytope of index @xmath0 . then @xmath693 equals @xmath722;u , v),\ ] ] where the sum is over all pairs of faces @xmath723 , @xmath724 . there is an important identity on @xmath592-polynomials , called stanley s convolution , cf . cor.8.3 ) , that can be stated in the following form , see ( * ? ? ? * lemma 11.2 ) , @xmath725;1,v)$ ] equals @xmath116 , if @xmath726 , and @xmath121 , otherwise . as a corollary we get @xmath727 @xmath728 where @xmath729 is the lattice volume ( with convention @xmath730 ) . when @xmath101 is a three - dimensional reflexive polytope , theorem [ hodge ] below yields that @xmath731 equals the topological euler - characteristic of a @xmath698 surface . hence , @xmath732 . however , there is still no direct combinatorial proof for this equation , cf . @xcite . the main motivation for the definition of the @xmath590-function and the calabi - yau dimension , and for the related conjectures stems from the following theorem : ( * ? ? ? * thm.7.2 ) let @xmath150 be a @xmath67-dimensional reflexive gorenstein cone of index @xmath0 with support @xmath187 , where @xmath167 is a cayley polytope of @xmath0 lattice polytopes . let @xmath733 be a generic complete intersection calabi - yau defined by these @xmath0 equations , hence @xmath734 . then @xmath735 is a polynomial in @xmath677 , called _ @xmath590-polynomial _ of @xmath733 , where the coefficients @xmath736 are the so - called _ stringy hodge numbers _ of @xmath733 , in the sense of @xcite . [ hodge ] this is a simplified version of the main result in @xcite . note , that the combinatorial identity ( [ comby ] ) gives the topological euler - characteristic of the calabi - yau @xmath733 . from this theorem we can derive the following result using algebraic geometry ( for @xmath737 , @xmath733 is a @xmath698-surface ) and the reciprocity - law ( remark [ erez](2 ) ) : conjecture [ e - conj ] holds , if at least one of the two gorenstein polytopes @xmath101 or @xmath118 of index @xmath0 is a cayley polytope of length @xmath0 . here is an example for a gorenstein polytope @xmath101 that does not satisfy the assumption of the previous corollary . let @xmath738 , where @xmath447 is the unimodular @xmath78-simplex , and let @xmath739 be the reflexive gorenstein cone of index two with support @xmath444 . we also define @xmath740 , and @xmath741{}^{{\scriptstyle\vee}}}$ ] . then by lemma [ directsum ] below , the support @xmath101 of the direct sum @xmath742 is a @xmath743-dimensional gorenstein polytope of index @xmath218 . the dual gorenstein polytope @xmath118 is the support of @xmath744 . since @xmath443 does not contain a special @xmath121-simplex , both @xmath101 and @xmath118 do not contain special @xmath78-simplices , hence by prop . [ ath ] both gorenstein polytopes of index @xmath218 are not cayley polytopes of length @xmath218 . the proof of the following lemma is left to the reader . let @xmath745 , @xmath746 be two reflexive gorenstein cones of indices @xmath747 , respectively @xmath748 . then the direct sum @xmath749 is a reflexive gorenstein cone of index @xmath750 , with dual cone @xmath751{}^{{\scriptstyle\vee}}}\oplus \sigma_2^{\makebox[0mm]{}^{{\scriptstyle\vee}}}$ ] . moreover , the support of @xmath742 contains a special @xmath752-simplex if and only if the support of @xmath739 contains a special @xmath753-simplex and the support of @xmath754 contains a special @xmath755-simplex . [ directsum ] if in theorem [ hodge ] the cone @xmath150 is associated to a nef - partition , then we get from remark [ erez](2 ) for the @xmath590-polynomial of @xmath733 @xmath756 where @xmath757 is a @xmath40-dimensional complete intersection calabi - yau defined by the dual nef - partition . in particular , stringy hodge numbers satisfy the following mirror symmetry property : @xmath758 it is still an open question whether there is only a finite number of topologically different calabi - yau @xmath675-folds ( @xmath759 ) . we can look at this question from a combinatorial point of view and ask another one : let @xmath675 be fixed . is there up to scalar only a finite number of polynomials in @xmath677 that occur as @xmath590-polynomials of gorenstein polytopes @xmath101 such that @xmath694 equals a fixed number @xmath675 ? [ conj ] in particular , a positive answer would imply that the set of stringy hodge numbers associated to irreducible calabi - yau complete intersections @xmath733 defined by a nef - partition is bounded . this implication follows from the fact that by poincar duality the leading coefficient @xmath156 of @xmath760 equals the constant coefficient which is @xmath121 , because @xmath761 is the number of irreducible components of @xmath733 . we observe that in general the number @xmath156 can be arbitrary large powers of @xmath134 . indeed , let @xmath762 be a nef - partition of length @xmath0 and let @xmath665 be @xmath590-polynomial of the corresponding calabi - yau complete intersection @xmath733 of dimension @xmath763 . then @xmath764 polytopes @xmath765 ) \subseteq \mr \oplus \r\ ] ] define a reducible nef - partition of length @xmath764 of the same cy - dimension @xmath763 . the corresponding calabi - yau complete intersection @xmath766 is a product of @xmath733 with two points . therefore @xmath590-polynomial of @xmath766 equals @xmath767 . by repeating the same procedure , we get @xmath590-polynomials with arbitrary large leading coefficient @xmath156 . [ alggeonefpart ] there are various finiteness results on toric fano varieties @xcite . however these results are not sufficient to prove finiteness of @xmath590-polynomials of gorenstein polytopes of fixed calabi - yau dimension . we are still optimistic and want to illustrate our hope by an example of @xmath1-dimensional gorenstein simplices @xmath768 of index @xmath0 defined as @xmath769 where @xmath770 are positive integers having the property @xmath771 @xmath494 . let @xmath772 @xmath773 . the fact that @xmath768 is a gorenstein simplex of index @xmath0 is equivalent to the equation @xmath774 if @xmath775 for some @xmath274 , then @xmath768 is a lattice pyramid over a @xmath22-dimensional simplex and , by [ pyramid ] , the @xmath590-polynomial of @xmath768 is @xmath116 . therefore it is enough to consider only the case when @xmath776 @xmath773 . let @xmath777 be the number of indices @xmath274 such that @xmath778 . then we get from the above equation @xmath779 or @xmath780 . by definition @xmath781 so we obtain @xmath782 . this implies that if @xmath768 has a fixed calabi - yau dimension @xmath783 , then there exist only finitely many possibilities for integers @xmath784 which are different from @xmath134 . in order to compute @xmath785 we observe that the simplex @xmath768 determines a landau - ginzburg model with lg - potential of fermat type : @xmath786 the corresponding formula for the @xmath590-polynomial is expected to coincide with the well - known formula of vafa @xcite . the latter one does not change if one adds any quadratic terms to @xmath787 . therefore we come to finitely many @xmath590-polynomials for a fixed value of @xmath788 . let @xmath3 and @xmath4 be dual to each other gorenstein polytopes of index @xmath789 . example [ cube ] shows that @xmath3 and @xmath4 may contain many different special @xmath5-simplices . consider the case when a gorenstein polytope @xmath3 contains two different special @xmath5-simplices @xmath386 and @xmath790 . assume that the dual gorenstein polytope @xmath4 also contains a special @xmath791-simplex @xmath362 . then the special @xmath5-simplex @xmath362 defines a cayley structure on @xmath3 , i.e. , some lattice polytopes @xmath61 such that @xmath792 . now two different special @xmath5-simplices @xmath793 define two collections of lattice points @xmath794 ( @xmath795 such that @xmath796 is the unique interior lattice point in the reflexive polytope @xmath293 . we remark that two cayley structures of the dual gorenstein polytope @xmath4 corresponding to @xmath386 and @xmath790 may define very different polytopes @xmath513 . we consider below a picture of two centered nef - partitions corresponding to polytopes @xmath797 and @xmath798 which are the same up to a affine translation , i.e. @xmath799 and @xmath800 : [ fig2 ] we see that the corresponding polytopes @xmath801 and @xmath802 are not isomorphic via affine translations by lattice vectors . it is easy to see that , if two nef - partitions consist of collections of lattice polytopes @xmath803 and @xmath804 such that @xmath805 for @xmath806 @xmath501 , then the corresponding generic calabi - yau complete intersections @xmath807 and @xmath808 are isomorphic . however , there is no natural isomorphism between the generic mirror calabi - yau complete intersections @xmath809 and @xmath810 . using the formula for @xmath91-polynomial of @xmath4 , we can only say that the stringy hodge numbers of @xmath809 and @xmath810 are the same . are the mirror calabi - yau complete intersections @xmath809 and @xmath810 birationally isomorphic ? even if in general the answer to this question will be negative , we can use the approach of the homological mirror symmetry and formulate the following conjecture : there exists a equivalence ( of fourier - mukai type ) between the derived category of coherent sheaves on two calabi - yau complete intersections @xmath809 and @xmath810 . this conjecture needs some comments , because calabi - yau varieties @xmath809 and @xmath810 are usually singular . however , using toric methods , one can always construct their partial crepant desingularizations @xmath811 and @xmath812 which are calabi - yau varieties having at worst terminal abelian quotient singularities . as in the toroidal case considered by kawamata in @xcite the equivalence class of stacky derived categories of @xmath811 and @xmath812 should be independent of the choice of the partial crepant desingularizations . so we expect the existence of a fourier - mukai transform which establishes an equivalence between the stacky derived categories of @xmath811 and @xmath812 . in this concluding section we give as an addendum to section 3 a concise survey on how to modify , project and decompose nef - partitions . by collecting lattice polytopes in a nef - partition we obtain another nef - partition . for this , we simply replace some lattice polytopes in a nef - partition by their minkowski sum . this yields a shorter nef - partition , which is by cor . [ nabcor ] dual to the nef - partition we get by replacing the respective lattice polytopes in the dual nef - partition by their convex hull : let us explain how prop . [ regroup ] relates to the results of section 2 : more generally , let @xmath61 be arbitrary lattice polytopes with @xmath293 reflexive . then by theorem [ sum ] , @xmath824 is a gorenstein cayley polytope of index @xmath0 . the dual gorenstein polytope @xmath4 contains a special @xmath5-simplex with vertices @xmath825 ( see prop . [ ath ] ) . theorem [ brtheo ] shows that projecting @xmath4 along the affine hull of this special simplex yields the reflexive polytope @xmath826 . now , let @xmath818 be a partition of @xmath819 into non - empty subsets . we may assume @xmath827 . in the same way as in the proof of theorem [ sum ] and theorem [ brtheo ] , projecting @xmath4 along @xmath828 yields a gorenstein polytope dual to the gorenstein polytope @xmath829 of index @xmath830 . after @xmath831 such projection steps , we finally get a gorenstein polytope of index @xmath831 dual to @xmath832 . in particular , when @xmath61 and @xmath833 are dual to each other centered nef - partitions , we see that @xmath834 , and the projection along @xmath835 yields @xmath836 as the gorenstein polytope dual to @xmath837 . hence , eventually we get @xmath838 as the gorenstein polytope of index @xmath831 dual @xmath832 . by translation we may assume @xmath845 . let @xmath418 be a facet of @xmath843 . then the preimage @xmath124 of @xmath418 is a minkowski sum @xmath846 for a face @xmath847 of @xmath62 . since @xmath848 , we easily see that @xmath849 . now , the statement follows from @xmath850 . we may suppose @xmath841 . let @xmath53 be the unique interior lattice point of @xmath844 . by considering @xmath851 instead of @xmath840 we may assume that @xmath844 is reflexive with respect to @xmath116 . let @xmath81 be any lattice point of @xmath3 . then by considering @xmath852 instead of @xmath840 we may also assume that @xmath853 . hence @xmath854 . so @xmath843 has @xmath116 as an interior lattice point . now , lemma [ minky ] yields that any vertex of @xmath855 is also a vertex of @xmath856 . since @xmath856 is a lattice polytope , @xmath857 is also a lattice polytope . let @xmath871 such that @xmath817 is a reflexive polytope ( in its linear span ) . by prop . [ projtheo ] it is enough to assume that @xmath817 contains @xmath116 in its relative interior . we say @xmath872 is _ irreducible _ , if @xmath872 is minimal with this property . theorem 5.8 of @xcite states that any ( proper ) nef - partition has a unique _ decomposition _ into irreducible ( proper ) nef - partitions , i.e. , there are unique irreducible subsets @xmath818 that form a partition of @xmath819 such that @xmath873 this is called _ semi - simplicity of nef - partitions_. here , we recall the proof of the existence of such a decomposition and give a generalization in prop . [ cancel ] . we also prove in prop . [ unique ] the uniqueness of this decomposition , which was missing in @xcite . finally , we determine the maximal length of a proper nef - partition in prop . [ length ] . the existence of a decomposition into irreducible nef - partitions is an immediate consequence of the following result , which was shown in ( * ? ? ? * thm.5.8 ) ( we include a proof for the convenience of the reader ) : by prop . [ regroup ] @xmath876 and @xmath877 are dual nef - partitions . since @xmath817 contains @xmath116 in its relative interior , prop . [ projtheo ] yields @xmath878 . moreover , @xmath879{}^{{\scriptstyle\vee}}}$ ] because of @xmath880 . furthermore , @xmath881 , since @xmath882 is only possibly for @xmath883 . we show that @xmath150 is actually a linear subspace of @xmath100 . this already yields @xmath884 , and by projecting along @xmath885 we get that @xmath874 is reflexive by prop . [ projtheo ] . let @xmath886 . we have to show @xmath887 . let @xmath198 be the projection along @xmath888 . as @xmath116 is in the interior of @xmath889 , and @xmath890 , we get @xmath891 . therefore there is some @xmath892 with @xmath893 . this implies @xmath894 , thus @xmath895 . by prop . [ regroup ] , @xmath901 and @xmath877 are dual nef - partitions . the same is true for @xmath902 and @xmath903 . hence , prop . [ projtheo ] , implies that if some of these @xmath904 polytopes contains @xmath116 in its relative interior , then it is necessarily a reflexive polytope . since @xmath817 has @xmath116 in its relative interior if and only if @xmath816 has @xmath116 in its relative interior , we may assume that @xmath817 is reflexive . [ exprop ] yields that also @xmath874 is reflexive . now , cor . [ suppfct ] implies @xmath905{}^{{\scriptstyle\vee}}}\,:\ , { \langle \delta^i , y \rangle } \geq -1\ } = \{y \in ( \delta_j)^\perp \,:\ , { \langle \delta^i , y \rangle } \geq -1\},\ ] ] since @xmath906 has @xmath116 in its relative interior . from this , we get @xmath907 . hence , @xmath864 is reflexive . in the same way we see @xmath908 , so also @xmath909 is reflexive . now , the last statement follows merely from the definition of dual nef - partitions . let @xmath840 be lattice polytopes with @xmath844 reflexive . if @xmath3 contains a lattice point in its relative interior , then @xmath3 and @xmath62 are reflexive polytopes , and @xmath840 form a nef - partition . [ cancel ] we define @xmath910 . since @xmath911 , there exist @xmath912 with @xmath913 . hence @xmath914 . assume there is some @xmath915 , @xmath916 . since @xmath116 is in the relative interior of @xmath3 , there is a linear form @xmath564 in the dual space of @xmath917 such that @xmath918 , and @xmath919 is a facet of @xmath3 . let @xmath920 with @xmath921 . we have @xmath922 . hence , @xmath923 is a face of @xmath924 . since @xmath925 , @xmath926 and @xmath927 , we see that @xmath418 is a facet of @xmath928 , parallel to @xmath124 . since @xmath844 is reflexive , there is an integral linearform @xmath929 in the dual space of @xmath917 such that @xmath930 . we have @xmath931 for @xmath932 . let @xmath205 be a vertex of @xmath124 , hence @xmath933 . this implies @xmath934 , a contradiction . let @xmath937 and @xmath938 . then @xmath844 has only one interior lattice point , @xmath3 is reflexive , however @xmath62 has no interior lattice point . this example , due to christian haase , shows that in prop . [ cancel ] the assumption that @xmath844 is reflexive can not be weakened by the condition that @xmath844 has only one interior lattice point . [ without ] by prop . [ exprop ] we have @xmath947 , thus @xmath948 . hence , since @xmath949 contains @xmath116 in its relative interior , also @xmath950 contains @xmath116 in its relative interior . this contradicts @xmath872 being irreducible . an irreducible proper nef - partition has length @xmath952 , see ( * ? ? ? * cor.3.5 ) or ( * ? ? ? * lemma 2.6 ) . now , a proper nef - partition of length @xmath0 decomposes into , say @xmath831 , irreducible nef - partitions . this yields @xmath953 where complete equality holds if only if any irreducible nef - partition has dimension one and length two . batyrev , _ stringy hodge numbers of varieties with gorenstein canonical singularities _ , in : m .- h . saito ( ed . ) et al . , integrable systems and algebraic geometry , proceedings of the 41st taniguchi symposium , japan 1997 , singapore : world scientific , 1 - 32 , 1998 . batyrev , _ lattice polytopes with a given @xmath397-polynomial _ , in : c.a . athanasiadis ( ed . ) et al . , algebraic and geometric combinatorics , proceedings of a euroconference in mathematics , anogia , crete , greece , august 2026 , 2005 , providence , ri : ams , contemp . math . * 423 * , 110 , 2007 . batyrev & l.a . borisov , _ on calabi - yau complete intersections in toric varieties _ , in : m. andreatta ( ed . ) et al . , higher dimensional complex varieties , proceedings of the international conference , trento , italy , june 1524 , 1994 , berlin : walter de gruyter , 3965 , 1996 . batyrev & l.a . borisov , _ dual cones and mirror symmetry for generalized calabi - yau manifolds _ , in : b. greene ( ed . ) et al . , mirror symmetry ii , cambridge , ma : international press , ams / ip stud . adv . math . * 1 * , 7186 , 1997 . emiris & e.p . tsigaridas , _ minkowski decomposition of convex lattice polygons _ , in : m. elkadi , b. mourrain , r. piene ( eds . ) , algebraic geometry and geometric modeling , berlin : springer - verlag , 2005 . c. haase , _ reflexive polytopes in dimension @xmath134 and @xmath78 , and the numbers @xmath954 and @xmath692 _ , in : m. beck , b. chen , l. fukshansky , c. haase , a. knutson , b. reznick , s. robins , and a. schrmann , problems from the cottonwood room , contemporary mathematics * 374 * ( 2005 ) , 179191 . r. stanley , _ generalized h - vectors , intersection cohomology of toric varieties , and related results _ , in : commutative algebra and combinatorics , us - jap . joint semin . , kyoto / jap . 1985 , adv . stud . pure math . * 11 * ( 1987 ) , 187213 . j. stienstra , _ resonant hypergeometric systems and mirror symmetry _ , in : m .- h . saito ( ed . ) et al . , integrable systems and algebraic geometry , proceedings of the 41st taniguchi symposium , kobe , japan , june 30-july 4 , 1997 , and in kyoto , japan , july 7 - 11 , 1997 , singapore : world scientific , 412 - 452 , 1998 .	the purpose of this paper is to review some combinatorial ideas behind the mirror symmetry for calabi - yau hypersurfaces and complete intersections in gorenstein toric fano varieties . we suggest as a basic combinatorial object the notion of a gorenstein polytope of index @xmath0 . a natural combinatorial duality for @xmath1-dimensional gorenstein polytopes of index @xmath0 extends the well - known polar duality for reflexive polytopes ( case @xmath2 ) . we consider the borisov duality between two nef - partitions as a duality between two gorenstein polytopes @xmath3 and @xmath4 of index @xmath0 together with selected special @xmath5-dimensional simplices @xmath6 and @xmath7 . different choices of these simplices suggest an interesting relation to homological mirror symmetry .
the photon induced reactions on nucleons where mesons and baryons are measured as final decay products , have revealed a rich excitation spectrum of nucleons and other baryons that reflects their complicated multi - quark inner dynamics . the determination of the properties of the nucleon excited states ( or resonances ) from the data of such experiments is a major challenge of hadron physics @xcite . the effective methods that are used to describe the dynamics of meson production reactions , include explicitly the baryon resonance states whose properties are extracted by comparing the predictions of the theory with the experimental data @xcite . however , for a reliable realization of this task , a model is required that can analyze different reactions over the entire energy range using a single lagrangian density , which generates all non - resonance contributions from born , @xmath2- and @xmath3-channel contributions without introducing new parameters . at the same time , the lagrangian should also satisfy the symmetries of the fundamental theory ( i.e. qcd ) while retaining only mesons and baryons as effective degrees of freedom . the coupled - channels method within the @xmath4-matrix approximation is one model that respects these constraints while providing a way to analyze simultaneously all the reaction data for a multitude of observables in different reaction channels . it also leads to a convenient way of imposing the unitarity constraint because the @xmath5 matrix in this approach is unitary provided the @xmath4-matrix is taken to be real and hermitian ( see , e.g. , ref . @xcite ) . the experimental data on the @xmath6 meson and associated strangeness ( @xmath7 and @xmath8 ) production in photon - nucleon reactions are available for photon energies ( @xmath9 ) in the range of threshold to 3 gev . thus they cover not only the entire resonance region but also the region where the background contributions ( @xmath3-channel amplitudes mainly ) are expected to be dominant . the data for the @xmath6 channel are reviewed recently in ref . the coupled - channels @xmath4-matrix method has been used extensively in refs . @xcite to describe the data of both these channels in the entire range of @xmath9 with a single set of parameters . these reaction channels have also been studied within the giessen coupled - channels @xmath4-matrix model in refs . @xcite for @xmath9 between threshold to 2.0 gev . [ cols="^,^ " , ] in the left panel of fig . 3 , we show the comparison of the scelm calculations with the data for the @xmath10 of the @xmath11 reaction as a function of @xmath12 beam momentum ( @xmath13 ) . it is clear that the data are reproduced reasonably well by the model within the statistical errors . the zero degree differential cross sections for the hypernuclear production reactions @xmath14c@xmath15be and @xmath16si@xmath17 mg calculated by using the same vertex parameters , are shown in the right panel of fig . 3 . in these calculations we have employed pure single - particle - single - hole @xmath18 configurations to describe the nuclear structure part . the bound proton - hole and @xmath19-particle state spinors were generated in a phenomenological model , where they are obtained by solving the dirac equation with scalar and vector fields having woods - saxon radial forms . with a set of radius and diffuseness parameters , the depths of these fields are searched to reproduce the binding energy ( be ) of the given state . the bes for the @xmath19 bound states were taken to be those predicted in the latest version of the quark - meson coupling ( qmc ) model @xcite as their experimental values are still unknown .. in this figure , we have shown results for populating the hypernuclear state with maximum spin of natural parity for each configuration . we see that the zero degree differential cross sections for the @xmath19 hypernuclear production reactions peak around the beam momentum of 1.0 gev / c with peak cross section of more than 1 @xmath20 . they closely follow the trends of the elementary @xmath19 production cross sections . our predictions will be useful for the future jparc experiments . the heavy flavored hadrons consisting of relatively heavier mass charm quarks provide an additional handle for the understanding of qcd . the charm quark introduces a mass scale that is much larger than the confinement scale @xmath22 mev . in contrast , the energy scale of the lighter quarks is @xmath23 . the presence of two scales in such systems naturally leads to the construction of an effective theory where one can actually calculate a big portion of the relevant physics using perturbation theory and renormalization - group techniques . in this context , the investigations of the production of heavy flavor hadrons are of great interest . since the discovery of @xmath24 in 1974 @xcite , the production of charmonium ( @xmath25 ) states has been extensively studied in electron - positron and proton - antiproton ( @xmath21 ) annihilation experiments . yet a substantial part of the charmonium spectrum is still to be precisely measured . the first charmed baryon states were detected in 1975 in neutrino interactions @xcite . although many new excited charmed baryon states have been discovered since then , the studies of the production and spectroscopy of the charmed baryons have not been carried out in the same detail as the charmonium states . in the near future , charmed hadron production will be investigated in the @xmath26 annihilation in the ( @xmath27 ) experiment at the fair in gsi , darmstadt @xcite . the advantage of using antiprotons in the study of the charmed baryon is that in @xmath26 collisions the production of no extra particle is needed for the charm conservation , which reduces the threshold energy . moreover , all the the charmonium states can be reached directly in this method , which is in contrast to the electron - positron annihilation where direct formation is allowed for only those final charmonium states that have the quantum numbers of the photon ( @xmath28 ) . for the planning of these experiments at the @xmath27 facility , reliable theoretical estimates of the cross section of @xmath29 , @xmath30 and @xmath31 reactions are of crucial importance @xcite . in ref . @xcite , the scelm has been used to study the @xmath32 reaction as a sum of the @xmath3-channel @xmath33 and @xmath34 meson - exchange diagrams [ fig . a similar model is employed in ref . @xcite to calculate the cross sections of @xmath30 , and @xmath35 reactions as a sum of the @xmath3-channel @xmath36 and @xmath37 , and @xmath38 baryon - exchange diagrams , respectively [ figs . 4(b ) and 4(c ) ] . also included in these calculations are the @xmath39-channel excitation , propagation and decay of the @xmath40 resonance into the @xmath41 and @xmath42 channels . it is found that the @xmath10 of the @xmath43 and @xmath44 reactions at the @xmath45 beam momentum of 15 @xmath46 ( which the regime of the @xmath27 experiment ) are around 25 @xmath47b and 550 @xmath48b , respectively . these values are drastically larger than those predicted in previous calculations based on models that invoke the quark degrees of freedom . the future @xmath27 experiment will clarify the adequacy of various models at these relatively lower beam momenta . in * summary * , we can say that effective lagrangian models are able to describe reasonably well the intermediate energy nuclear reactions . the coupled - channels @xmath4-matrix method allows for the simultaneous calculation of the observables for a large multitude of reactions with considerably fewer parameters than would be necessary if each reaction channel were fitted separately . for the strangeness and charm production reactions , a single - channel effective lagrangian method with suitably chosen parameters is able to make important predictions that will be tested in the future experiments planned at upcoming facilities . * acknowledgments * : this work has been supported by the helmholtz international center ( hic ) for fair and the council of scientific and industrial research ( csir ) , india . the author acknowledges useful discussions with h. lenske , o. scholten , a.w . thomas and k. tsushima .	in the domain of nuclear reactions at intermediate energies , the qcd coupling constant @xmath0 is large enough ( @xmath1 0.3 - 0.5 ) to render the perturbative calculational techniques inapplicable . in this regime the quarks are confined into colorless hadrons and it is expected that effective field theories of hadron interactions via exchange of hadrons , provide useful tools to describe such reactions . in this contribution we discuss applications of one such theory , the effective lagrangian model , in describing the hadronic reactions at intermediate energies whose measurements are the focus of a vast international experimental program . in the quest for understanding the quantum chromodynamics ( qcd ) , the hadron - hadron , hadron - photon and hadron - lepton reactions have played a major role . from these studies two quite distinct regimes of qcd have emerged . the first regime is of those processes that involve very high momentum transfers . here the asymptotic freedom property of qcd has enabled the perturbative qcd ( pqcd ) methods to describe @xcite a large set of high energy large momentum transfer reaction observables with great precision , providing the most precise test of qcd to date @xcite . the second regime that includes nuclear reactions at intermediate energies , is characterized by the confinement of quarks inside the hadrons . this remarkable feature , along with the running of the qcd coupling constant ( @xmath0 ) towards larger values makes the pqcd methods inapplicable - one requires techniques that can solve the theory exactly . to date , the only fully predictive non - perturbative method for studying qcd at low energies is lattice qcd ( lqcd ) . the progress in lqcd studies of the low - lying spectrum of qcd , hadronic structure and interactions has been significant in the recent years ( see , eg . , ref . @xcite ) . nevertheless , lqcd calculations are still far from being routinely amiable to solutions of intermediate energy nuclear reactions . it is therefore , necessary to use effective methods to describe the dynamics of such processes . the effective field theories where hadrons interact via exchange of hadrons , provide a viable alternative to lqcd for the description of the intermediate energy nuclear reactions ( see , e.g. , refs . @xcite ) . these theories implement the symmetries of the underlying theory , qcd , by using proper effective lagrangians . however , many of these models invariably require _ a priory _ knowledge of unknown parameters related to the coupling constants associated with the effective lagrangians and the masses of the hadrons . because the effective theories provide flexible theoretical approaches that allow a swift understanding of the intricacies involved in the scattering problems at intermediate energies and have the ability to guide to further experimental studies , steady progress has been made towards determining the realistic input parameters involved in these theories . in this presentation we discuss applications of effective lagrangian based theoretical models , to describe the meson production reactions ( both non - strange and strange ) in photon and meson induced reactions on nucleons at intermediate energies . we also present some results on the charmed baryon and meson production in the antiproton - proton annihilation process .
the formation and evolution of the dark matter halo which can be treated as the self - gravitational collisionless stellar system have become a challenging issue in the study of dark matter . thanks to the improved computational power , n - body simulations of dark matter halos become more and more accurate and important with increasing resolution . n - body simulations such as the universal nfw profile@xcite and others@xcite show that the density profiles of dark matter halos behave as @xmath6 , where @xmath1 in the center and @xmath2 in the outer parts of halos . however , the numerical inner behaviors of dark matter halos are not supported by observations.@xcite some work indicates that density profiles of dark matter halos might become shallower than @xmath7 in the center@xcite , and perhaps even tend to be a core with no cusp at all.@xcite n - body simulations not only provide us the density profiles of halos , but also give the relevant information of the velocity space of collisionless particles in the halos . velocity dispersion and anisotropy profiles@xcite have been well described by simple analytical fits . two interesting phenomena from simulations indicate that velocity dispersion and the density profiles of the haloes are not independent . first , hansen and moore@xcite found that the density logarithmic slope @xmath8 is correlated to the velocity anisotropy which is parameterized by the anisotropy parameter @xmath9 and they provided the empirical formula @xmath10 . therefore @xmath3 ( isotropic velocity dispersion ) in the inner part as @xmath11 and @xmath12 in the outer part as @xmath13 . second , it has been argued@xcite that the so called pseudo phase - space density follows a power law @xmath14 with exponent @xmath15 , where @xmath16 is the density profile and @xmath17 is the total velocity dispersion . subsequent studies have confirmed that @xmath18 is a power law in radius , but the best fitting values of the exponent @xmath19 diverse from each other@xcite and range from @xmath20 to @xmath21 . because @xmath4 has the same dimension as the phase - space density , @xmath4 has been called pseudo phase - space density or `` poor - man s '' phase - space density@xcite . with these results in mind , much theoretical work has been done for the study of the relation between density profile behavior and pseudo phase - space density . some authors examine this matter by solving the jeans equation . williams et al got a critical exponents@xcite @xmath22 and dehnen and mclaughlin calculated corresponding density profiles in both isotropic and anisotropic cases@xcite . some other work solved the jeans equation to explore the relation between density profile and pseudo phase - space density @xcite . recently , r. n. henriksen@xcite considered a series expansion for a dark matter distribution function in the spherically symmetric anisotropic limit to discuss pseudo phase - space density . in this paper , we concentrate on the center of the dark matter halo where the velocity dispersion is almost isotropic and calculate the distribution function and pseudo phase - space density where @xmath23 in the spherically symmetric case . in section 2 we review the basic knowledge about distribution function which is needed for this paper . in section 3 , the distribution function , velocity dispersion and pseudo phase - space density in the center of the dark matter halo are calculated with the given density profile . we make the discussion and conclusion in section 4 . in the center of dark matter halos , with spherically symmetric assumption , the distribution function @xmath24 can be reduced to @xmath25 as anisotropy parameter @xmath3 ( almost isotropic ) when @xmath23 . if we know the distribution function , we can calculate the density profile and total velocity dispersion profile as below:@xcite , @xmath26 @xmath27 where the binding energy @xmath28 is defined as @xmath29 and @xmath30 is the relative gravitational potential which can be obtained from the poisson equation : @xmath31 @xmath32 is the radial velocity and @xmath33 is the tangential velocity : @xmath34 eddington@xcite provided the inversion formula of eq . ( [ rhoiso ] ) : @xmath35 where @xmath36 denotes the 2nd order differentiation operator with respect to @xmath28 . the anisotropy parameter@xcite mentioned in the introduction is defined as : @xmath37 where @xmath38 and @xmath39 are the tangential and radial velocity dispersion . if @xmath40 , the velocity dispersion is tangentially anisotropic . @xmath41 correspond to the radially anisotropic cases , and @xmath42 , the isotropic case . in this section , we will calculate the distribution function and pseudo phase - space density with the given density profile in the very central region of the spherically symmetric dark matter halo where the velocity dispersion is almost isotropic . let us consider a family of density profiles with parameters @xmath43 and @xmath44 named as new generalized nfw profiles,@xcite @xmath45 we set the characteristic radius @xmath46 , the total mass @xmath47 and the gravitational constant @xmath48 here . then the density profiles reduce to @xmath49 @xmath50 then the relative potential @xmath30 can be calculated from the poisson eq . ( [ eq : poisson]).@xcite @xmath51 \\ & a_2(r)=\ _ 2\tilde{f}_1[2-a+\epsilon,\ -a,\ 2-a,\ -r ] \\ & a_3(r)=\ _ 2\tilde{f}_1[3-a+\epsilon,\ -a,\ 2-a,\ -r ] \label{eq : psi } \end{split}\end{gathered}\ ] ] where @xmath52 is the relative gravitational potential at @xmath53 which is determined by the condition @xmath54 . and then , we calculate the asymptotic approximation of @xmath55 and @xmath56 as below : @xmath57 @xmath58 from @xmath55 and @xmath56 , we can get the asymptotic approximation of @xmath59 when using eq . ( [ fiso ] ) @xmath60 } { 4\sqrt{2}(2-a)^2 e^{3/2}(e-\psi_0)^2 \pi^2 } \\ & \\ & a_4(e)=\ _ 2f_1[\frac{1}{2},\ \frac{a}{2-a},\ \frac{3}{2},\ -\frac{e}{\psi_0-e } ] . \label{fe } \end{split}\end{gathered}\ ] ] we should note that this result of distribution function is only valid in the very center of dark matter halo where the velocity dispersion is almost isotropic . in the outer part of the halo , no matter whether the velocity dispersion is isotropic or not , this result is invalid and even unphysical . it s easy to note that , when @xmath23 , the binding energy @xmath28 approaches to its maximum value @xmath61 . following this observation , we can further reduce eq . ( [ fe ] ) in the parameter space @xmath62 which has covered the result of simulations . nevertheless , it s just in this parameter scope that f(e ) in eq . ( [ fe ] ) is physically meaningful : @xmath63 by using equations ( [ sigma ] ) , ( [ rhopsi ] ) and ( [ limitfe ] ) , we can get the total velocity dispersion in the very center of the halo , @xmath64 , \label{limitsigma } \\ & \\ & \begin{array}{rcl } \left._2f_1[\frac{5}{2},\ \frac{1}{2}-\frac{2}{a-2},\ \frac{7}{2},\ \frac{\psi}{\psi-\psi_0}]\right\|_{\psi\rightarrow \psi_0}&=&\frac{5(a-2)}{4(a-1)}(\frac{\psi}{\psi_0-\psi})^{-\frac{1}{2}+\frac{2}{a-2 } } \\ \\ & & + \frac{15\sqrt{\pi}\gamma(-2-\frac{2}{a-2})}{8\gamma(\frac{1}{2}-\frac{2}{a-2})}(\frac{\psi}{\psi_0-\psi})^{-5/2}. \end{array } \end{split}\end{gathered}\ ] ] then , it s not difficult to simplify equation ( [ limitsigma ] ) to the final result : @xmath65 and substitute equation ( [ limitpsir ] ) into equation ( [ sigma2 ] ) to transform @xmath66 into @xmath67 @xmath68 now , after we know the total velocity dispersion and the density profile in the very center of the halo , the pseudo phase - space density can be calculated directly , @xmath69 the power - law behaviour of the pseudo phase - space density was first found by simulation@xcite and analyzed by subsequent theoretic work@xcite , but its physical meaning is still unclear . for this consideration , let us compare the distribution function and pseudo phase - space density in the very center of the halo . we should note that the velocity of particles , which may stably exist in the very center of the halo , approaches to @xmath70 with @xmath23 . if the velocity term of the binding energy eq . ( [ e ] ) or the total velocity dispersion @xmath17 is much smaller than the second term of the rhs of eq . ( [ limitpsir ] ) or at most has the same asymptotic behavior as the second term of rhs of eq . ( [ limitpsir ] ) , this condition requires @xmath71 which can be obtained from eq . ( [ limitpsir ] ) and eq . ( [ sigma : r2 ] ) , then @xmath72 . with this formula and eq . ( [ limitfe ] ) , we can get @xmath73 from eq . ( [ gamma2 ] ) and eq . ( [ fr ] ) , we see that , if @xmath74 , the distribution function and the pseudo phase - space density have the same asymptotic behavior where @xmath23 . this result indicates that the `` poor - man s '' phase - space density is not so poor but might relate to real phase - space density at least in the very center of the halo . n - body simulations show us some properties of the dark matter halo , such as density profile @xmath16 , anisotropy parameter @xmath9 , pseudo phase - space density , and so on . although the simulation s resolution is limited , these information should be considered properly in the theoretic analysis in order to make theoretic work more realistic . in this paper , we investigate the problem of the center of the halo based on the following facts : ( 1 ) the existence of universal density profiles such as nfw@xcite , moore99@xcite , and in particular the new generalized nfw profile used in our work ; ( 2 ) anisotropy parameter @xmath3 in the center of the halo ; ( 3 ) the pseudo phase space density follows a power law in radius @xmath5 ; ( 4 ) limited resolution in the central region . we should notice that in this paper all results are only valid in the very center of the halo . from a given density profile , we calculate the asymptotic approximation of the distribution function in the very center of the halo . then the total velocity dispersion and the pseudo phase - space density are obtained from the distribution function . ( [ gamma2 ] ) shows the relation between the two parameters @xmath19 and @xmath43 . if we set @xmath75 as some simulations indicate , then @xmath43 should be 0.776 which is smaller than nfw s result . otherwise , if we set @xmath76 to adapt nfw profile and moore s profile , then @xmath77 which is larger than the simulations result . this apparent contradiction between theoretic result and simulation is not so weird since the resolution in the central region is nt high enough to describe the very central region of the dark matter halo . comparing the distribution function ( real phase - space density ) with the pseudo phase - space density , we find that they have the same asymptotic behavior if @xmath78 . this result indicates that the distribution function and pseudo phase - space density are closely related , even though the assertion is rather premature that the distribution function might have a power - law behavior just as the pseudo phase - space density in the whole range of radius @xmath5 . this interesting suggestion may also give us more confidence in the study of pseudo phase - space density and some new clues to the construction of the distribution function . by the way , some authors@xcite also study the pseudo phase - space density in from @xmath79 . it is obvious that our conclusion in this paper is well - founded only if @xmath80 . so , from this point of view , @xmath4 may be the best choice of the pseudo phase - space density . for the intensive study of the dark matter halo , the full range of radius @xmath5 should be involved . we need to consider how to extend the relationship between `` real '' and `` pseudo '' phase - space density from the very central region to other regions where the velocity dispersion is anisotropic . furthermore , reasonable knowledge and new strategies should be introduced and developed for the construction of more realistic models . first , it is admitted that dark matter halos in the real world are not spherically symmetric . second , the physics in the center of the halo may be very complicated . third , the halo actually is a polycomponent system which contains the dark matter , baryonic matter and even a supermassive black hole in the center . it is not radical to say that new methods and ideas are still needed in the future research on this matter . we are grateful for an anonymous referee for his / her helpful and constructive comments to improve the manuscript . this work is supported by the scientific research foundation for the returned overseas chinese scholars , state education ministry of china , and by the chinese academy of sciences under grant no . kjcx3-syw - n2 . r. g. carlberg , h. k. c. yee , e. ellingson , s. l. morris , r. abraham , p. gravel , c. j. pritchet , t. smecker - hane , f. d. a. hartwick , j. e. hesser , j. b. hutchings and j. b. oke , _ astrophys . j. _ * 485 * ( 1997 ) l13 . l. l. r. williams , c. austin , e. barnes , a. babul and j. dalcanton , conference : `` baryons in dark matter halos '' . novigrad , croatia , 5 - 9 oct ( 2004 ) . eds : r. dettmar , u. klein and p. salucci , _ proceedings of science _ published by sissa , p.20.1 , astro - ph/0412442	n - body simulations of dark matter halos show that the density profiles of halos behave as @xmath0 , where the density logarithmic slope @xmath1 in the center and @xmath2 in the outer parts of halos . however , some observations are not in agreement with simulations in the very central region of halos . the simulations also show that velocity dispersion anisotropy parameter @xmath3 in the inner part of the halo and the so called `` pseudo phase - space density '' @xmath4 behaves as a power - law in radius @xmath5 . with these results in mind , we study the distribution function and the pseudo phase - space density @xmath4 of the center of dark matter halos and find that they are closely - related .
it is generally accepted that there is a pseudogap ( pg ) in the low energy excitation spectrum of hole - doped cuprate superconductors , with a definite energy scale ( @xmath10 ) that falls as the hole concentration @xmath8 is increased . however the details are controversial and a recent review @xcite gives equal weight to three possible scenarios : ( a ) @xmath11 falls to zero on the over - doped side together with the superconducting transition temperature @xmath12 , ( b ) @xmath11 falls sharply to zero for slightly over - doped samples with @xmath13 or ( c ) @xmath11 is similar to case ( b ) but there is no pg in the superconducting state . the @xmath8-dependences of the heat capacity @xcite , and the in - and out - of - plane penetration depths , @xmath14 and @xmath15 , of two grain - aligned single layer cuprates @xcite suggest that at low @xmath16 the condensation energy @xcite and the superfluid density @xcite fall abruptly below @xmath17 . this seems to support scenarios of type b , but the issue is still being debated @xcite . here we report ac susceptibility ( acs ) data @xcite showing how the in - plane superconducting penetration depth @xmath14 of bi@xmath0sr@xmath1cacu@xmath2o@xmath3 ( bi:2212 ) and bi@xmath4sr@xmath5ca@xmath6y@xmath7cu@xmath2o@xmath3 ( bi(y):2212 ) changes between @xmath8 = 0.08 and 0.21 . these are relatively straightforward measurements , on the same unaligned sample as a function of @xmath18 , that can easily be checked by other research groups . we believe that they also support scenarios of type b above . they also give a linear @xmath16 - dependence of @xmath19 , over a wide range of @xmath16 , possibly becoming `` super - linear '' for pure bi:2212 below 20 @xmath20 . combining our large values of @xmath21 at low @xmath8 with recent scanning tunnelling spectroscopy ( sts ) work @xcite , suggesting relatively large fermi arcs for @xmath22 = 20 and 45 @xmath20 , may also give important insights into cuprate superconductivity . .[table01]annealing conditions , hole concentrations and room temperature thermopower for bi:2212 and bi(y):2212 samples . for both powder and bulk samples eqn . [ p1-eq01 ] has been used to find @xmath8 from @xmath23 measured by acs , with @xmath24 @xmath20 for bi:2212 and @xmath25 @xmath20 for bi(y):2212 . the measured values of the room temperature thermoelectric power , s(290 ) are also shown , and give consistent values of @xmath8 . [ cols="<,<,^,^,^ " , ] xrd patterns for bi:2212 and bi(y):2212 on a logarithmic scale . impurity lines from bi@xmath2o@xmath26 or bi@xmath2o@xmath27 are marked by ( * ) and incommensurate satellite reflections @xcite by ( + ) . ( b ) @xmath8 is plotted _ vs. _ weight loss - lower scale and apparent oxygen loss - upper scale . heat treatment details are given in footnote . @xmath8 values were found from the acs @xmath22 of a sintered bi(y):2212 bar using eq . 1 and @xmath28 . , width=264,height=359 ] x - ray powder diffraction ( xrd ) patterns for the two samples in fig . [ ch03-fig01]a . show that the bi:2212 sample was phase pure to within the noise level of @xmath29 . two peaks that are not indexed arise from the incommensurate superstructure @xcite . they are also present for bi(y):2212 but this has two more peaks from 3 - 5% bi@xmath2o@xmath26 or bi@xmath2o@xmath27 . examples of how the oxygen content of the sintered and powder samples was varied by annealing in flowing gases and quenching into liquid nitrogen @xcite are given in table 1 . sample weights were measured immediately after quenching . @xmath8 values were obtained from the @xmath22 values measured by acs using the parabolic law @xcite : @xmath30 fig . [ ch03-fig01]b @xcite shows @xmath8 _ vs. _ cumulative weight changes for 11 annealing treatments on one of the sintered bi(y):2212 samples . after the initial irreversible changes ( steps 1 - 4 ) , the data are reversible and give a line of unit slope @xcite . this provides further experimental confirmation @xcite of the _ changes _ in @xmath8 during the quenching treatments , because oxygen in the sr - o bi - o reservoir layer is expected to be o@xmath31 and there are 2 cu atoms per formula unit . [ ch03-fig01]b also shows that the irreversible losses ( @xmath32 1 @xmath33 in 1.1 @xmath34 ) occurred continuously on heating from 450 to @xmath35 @xcite and do not affect the value of @xmath8 ( since the @xmath8 values at steps 1 and 5 are the same ) . loss of bi is the most likely cause since such a weight loss is equivalent to 0.004 bi per formula unit and for a valency bi@xmath36 this would only change @xmath8 by 0.008 - well within the error bars of points 1 and 5 in fig . [ ch03-fig01]b . fine powders were obtained by gently grinding 50 - 100 @xmath33 of the fully - oxygenated sintered material with a small pestle and mortar . two series of acs experiments on bi(y):2212 powders gave similar results , data shown in table 1 and figs . 2b and 3b are for the second set . after the final ( 13th ) quench , grain sizes were determined by measuring the dimensions of @xmath37 grains in a scanning electron microscope ( sem ) photograph . approximately 1/3 of the grains were not circular and for these the geometric mean of the two radii was used . for bi(y):2212 the grain radius ( @xmath38 ) at 50% cumulative volume ( cv ) was 1.5 @xmath39 , with @xmath38 = 0.53 and 2.20 @xmath39 at 10 and 90% cv respectively . for pure bi:2212 the powders were sedimented in acetone to remove large particles , giving @xmath40 , 0.31 and 1.01 @xmath39 for the 50 , 10 and 90% cv points respectively . vs. @xmath16 for ( a ) pure bi:2212 and ( b ) 15% y - doped bi:2212 . the @xmath8 values determined from @xmath23 are also shown.,width=264,height=377 ] acs measurements were made using standard techniques @xcite . in the limit where @xmath15 is much larger than both @xmath14 and the grain size , we can obtain @xmath14 since only the component of @xmath41 along the crystalline @xmath42 - axis will give rise to detectable diamagnetism . in this case the ac signal from randomly oriented grains is 1/3 of that for aligned grains with @xmath43 , because the average of @xmath44 over a sphere is 1/3 . therefore we simply multiply the observed ac signal by 3 and apply our usual analysis for aligned grains @xcite with @xmath43 . the acs data in figs . [ ch05-fig02]a and [ ch05-fig02]b are in emu/@xmath45 . multiplying by @xmath46 gives @xmath47 where the magnetic moment @xmath48 is reduced from @xmath49 for a perfectly diamagnetic sphere because of the finite value of @xmath14 . allowing for a finite value of @xmath15 has negligible effect @xcite . the acs signals in fig . [ ch05-fig02 ] were extrapolated linearly to zero to find @xmath23 giving @xmath24 @xmath20 for bi:2212 and @xmath25 @xmath20 for bi(y):2212 , and @xmath8 values were then found using eq . [ p1-eq01 ] . we note that the lower values of @xmath8 in table 1 were obtained by vacuum annealing which gives less control than using flowing gases . we believe that these samples are nevertheless uniformly doped for following reasons . ( i ) the samples were periodically re - oxygenated , e.g. between quenches q9 and q13 and q12 and q15 in table 1 , and the results showed that the vacuum anneals did not cause any irreversible changes . ( ii ) at @xmath50 oxygen diffusion in bi:2212 @xcite is so fast that the oxygen content in a 10 @xmath51 m diameter grain will attain uniformity in 1 second . ( iii ) the hole concentrations in table 1 obtained from @xmath22 for powders and sintered bars are very similar and agree with @xmath8 values determined from the room temperature thermoelectric power using the results of ref . . such good agreement would not be obtained if the oxygen content of the powder samples were substantially non - uniform . the raw acs data in fig . [ ch05-fig02 ] clearly show that the signal , i. e. @xmath14 , only changes strongly with @xmath8 on the under - doped side for @xmath52 and that for the bi:2212 sample it saturates for @xmath53 . also the well - defined onsets in the acs signals at @xmath23 are consistent with the finite size scaling analysis of specific heat data @xcite , that ruled out gross inhomogeneity at any doping level . for bi:2212 there is an increase in the slope of the diamagnetism below 20 @xmath20 for @xmath54 , but this is not visible for the y - doped samples over the range of @xmath8 values studied , nor for lightly zn - doped bi:2212 samples ( @xmath55 @xmath20 @xcite - data not shown here ) . ) plotted as a function of @xmath16 for ( a ) pure bi:2212 and ( b ) 15% y - doped bi:2212 . values of @xmath8 are the same as in fig . comparisons with weak coupling @xmath56-wave behavior for the highest and lowest values of @xmath8 are shown in the insets . results assuming 3:3:1 ellipsoids ( e ) with h @xmath57 to the short side , instead of spheres , are also shown in the upper inset . here @xmath21 values are factors of 1.23 and 1.1 larger than for spheres for @xmath8 = 0.207 and 0.105 respectively . , width=283 ] the data in figs . [ ch05-fig02]a and [ ch05-fig02]b have been analyzed in the usual way by summing london s expression @xmath58 for a superconducting sphere of radius @xmath59 over the measured particle size distribution and varying @xmath60 until @xmath47 equalled the measured value . the resulting values of @xmath14 are plotted as @xmath61 _ vs. _ @xmath16 in figs . [ fig.3]a and [ fig.3]b . for bi(y):2212 , @xmath61 _ vs. _ @xmath16 is linear from low @xmath16 up to @xmath22 for @xmath8 = 0.12 to 0.183 while for pure bi:2212 there is evidence for `` super - linear '' behavior . most of the data are slightly more linear than expected for a weak coupling bcs @xmath56-wave state , as shown in the insets to fig . [ fig.3 ] . a recent compilation based on various techniques @xcite suggests that @xmath62 is usually @xmath63 rather than the weak coupling value @xmath64 . so we would not expect large deviations from weak coupling @xmath56-wave , but , if anything , larger gaps near the nodes would cause a slower decrease in @xmath61 with @xmath16 at low @xmath16 . the @xmath16-dependence for hg:1223 @xcite is close to that for weak coupling @xmath56-wave , but a slightly faster , more linear decrease was observed for yba@xmath65cu@xmath66o@xmath67 @xcite and more recently for heavily under - doped yba@xmath65cu@xmath66o@xmath68 crystals @xcite . relatively few over - doped materials have been studied so the `` super - linear '' behavior could be a general property of clean over - doped cuprates . note that @xmath69 has only been reported for optimally - doped bi:2212 crystals @xcite and in fig . 3a the `` super - linear '' behavior is less marked in this case , i. e. for @xmath8 = 0.16 . however sem pictures @xcite of the sintered samples studied here showed that the crystallites of pure bi:2212 were especially thin and plate - like , @xmath70 0.3 @xmath39 thick , while those of bi(y):2212 were much thicker . it possible that this is could be partly responsible for the `` super - linear '' behavior , as indicated in the inset to fig . 3a , where results for an analysis based on ellipsoids with 3:3:1 aspect ratios are shown . so although the `` super - linear '' dependence is potentially an important result it needs to be confirmed by other measurements . and @xmath71 are plotted vs. @xmath8 . the red and blue dashed lines show data from heat capacity work @xcite . ( b ) plot of @xmath22 vs. @xmath72 , i.e. @xmath73 . the dashed line shows a linear variation while the solid line shows @xmath22 @xmath74 @xmath75 @xcite . , width=283 ] generally speaking @xmath76 is related to the low energy quasi - particle weight . if this were preferentially distributed near the nodes of the @xmath56-wave superconductor in @xmath77-space then @xmath76 would indeed rise at low temperature as observed . alternatively we note that the two chain cuprate , yba@xmath65cu@xmath78o@xmath79 shows a similar `` super - linear '' behavior which was ascribed to superconductivity induced in the chains by the proximity effect at low @xmath16 @xcite . this raises the intriguing possibility that the pairing interaction in bi:2212 is very @xmath77-dependent and that for over - doped bi:2212 samples , proximity coupling could be playing a role even for states within the cu - o@xmath65 planes . in fig . [ lambdap]a data for both samples show a strong linear increase in @xmath80 with @xmath8 for @xmath81 but @xmath80 suddenly becomes constant for @xmath82 . bearing in mind that @xmath80 is a measure of the superfluid density @xmath83 as @xmath84 this strongly supports scenario b for @xmath11 . as shown in fig . [ lambdap]b there is a region where @xmath22 increases linearly with @xmath83 , but the large intercept at @xmath22 = 40 @xmath20 means that the empirical uemura relation , @xmath22 @xmath74 @xmath83 , @xcite does not apply here . the solid line shows that at low @xmath8 our data are more compatible with @xmath22 @xmath74 @xmath75 found recently for yba@xmath65cu@xmath66o@xmath68 @xcite although for our bi:2212 samples the values of @xmath85 are much smaller for similar values of @xmath22 . the plot in fig . 4b is qualitatively consistent with the pg model of ref . but the initial increase of @xmath22 with @xmath83 in our data seems to be too sudden to give quantitative agreement . as also shown in fig . 4a the our direct measurements of @xmath21 are consistent with earlier results @xcite from a mean - field , ginzburg - landau ( gl ) analysis of the field - dependent heat capacity , except at low @xmath8 where we find smaller values of @xmath83 . this is probably because of the large error bars in @xmath83 obtained from the heat capacity at low @xmath8 where the superconducting contribution is small@xcite . in fig . 4a , the deviation of the two bi(y):2212 points at @xmath8 = 0.121 and 0.129 from the general trend of @xmath80 with @xmath8 might be connected with a 1/8th plateau effect @xcite in @xmath23 . however simple modification of the @xmath86 line near @xmath8 = 0.125 will not account for these two anomalous points . finally we briefly compare our results for @xmath83 with recent sts data.@xcite @xmath83 is a fermi surface property and if the density of states @xmath87 in the normal state is strongly energy ( @xmath88 ) dependent it is expected @xcite to be given by : @xmath89 where @xmath90 is the projection of the fermi velocity along the supercurrent direction and the average is taken over an energy range of the order of the superconducting gap . in a bcs - like @xmath56-wave situation this energy range will be @xmath91 where the product of the bcs parameters @xmath92 and @xmath93 is finite . the sts work suggests that for heavily underdoped bi:2212 samples with @xmath22 values of 20 and 45 @xmath20 there are fermi arcs ( strictly speaking in the superconducting state these are bogoliubov arcs ) whose length in * k-space is still @xmath94 of that of the large hole - like fermi surface seen in overdoped samples where there is no pseudogap . as the mott insulating state is approached both techniques show a loss in density of states or spectral weight near @xmath95 . but our results suggest that @xmath83 is reduced by a factor of 25 when @xmath22 = 40 @xmath20 while at first sight the sts work only gives a factor of 3 reduction . our data can be reconciled with sts if this loss of states also occurs in the region of the arcs , but over an energy range smaller than @xmath96 , so that the bogoliubov quasi - particle peaks are still visible in sts . this is qualitatively consistent with the model in ref . in which the pg is usually smaller than the superconducting gap . however we note that a microscopic theory of this effect also needs to consider what happens when a supercurrent is produced by uniformly displacing the fermi surface in k-space . assuming that the pg is not displaced , then this continuity requirement seems to imply that only regions with a finite density of states at , and close to , @xmath95 will contribute to the supercurrent . we report direct evidence that at low @xmath16 the superfluid density of bi:2212 falls rapidly for @xmath9 , which supports scenario b for the pseudogap in this widely - studied @xcite compound . for both samples @xmath80 is extremely small in the heavily under - doped region , down by a factor @xmath97 while @xmath23 remains relatively high ( @xmath98 40 @xmath20 ) . in conjunction with recent sts work @xcite this could be a useful constraint for theoretical models . for many values of @xmath8 there is evidence for a linear @xmath16-dependence of the superfluid density over an unusually wide range of @xmath16 and preliminary evidence for `` super - linear '' behavior below @xmath99 @xmath20 in over - doped samples of pure bi:2212 . we acknowledge funding from the epsrc ( u.k . ) , for the experimental facilities in cambridge , and further financial support from the royal society , euryi , and mext - ct-2006 - 039047 . w. a. was supported by the development and promotion of science and technology talents project ( d.p.s.t . ) , thailand . we thank j. w. loram , s. h. naqib , j. l. tallon and e. m. tunnicliffe for helpful discussions . 99 s. h@xmath100fner , m. a. hossain , a. damascelli and g.a . sawatzky , rep . prog . phys . 71 * , 062501 ( 2008 ) . p. w. anderson , physica c * 460 - 462 * , 3 ( 2007 ) . fischer , m . kugler , i . maggio - aprile and c . berthod , , rev . phys . * 79 * , 353 ( 2007 ) . j. w. loram , j. luo , j. r. cooper , w. y. liang and j. l. tallon , j. phys . solids , * 62 * , 59 ( 2001 ) . c. panagopoulos _ et al . b * 60 * , 14617 ( 1999 ) . c. panagopoulos , t. xiang , w. anukool , j. r. cooper , y. s. wang and c. w. chu , phys . b * 67 * , 220502(r ) ( 2003 ) . w. anukool , phd thesis , university of cambridge , u.k . july 2003 . y. kohsaka _ et al . _ , nature * 454 * , 1072 ( 2008 ) . a. mironov _ et al . _ , jcpds - international center for diffraction data , 46 - 0431 ( 1998 ) . to prevent water vapor condensing on the samples after the quench , the gold sample container was placed in a glass beaker and warmed to room temperature by a strong stream of flowing gas . this procedure also worked well for the powder samples that were contained in a gold foil envelope but typically 1 - 2 @xmath33 were lost each time . all samples were periodically overdoped and re - measured to check that there had been no preferential loss of smaller or larger particles during the quench . the sequence is composed of 4 treatments in o@xmath65 from 450 to @xmath35 , then @xmath101 in o@xmath65 ( 5 ) and again at step 8 . treatments 6 , 7 , 9 and 10 correspond to various atmospheres , including vacuum , at temperatures up to @xmath50 , while 11 represents slow cooling from 450 to @xmath102 in o@xmath65 . after the initial heating to @xmath35 ( steps 1 - 4 ) the data lie on the same reversible line . when referred to the oxygen content derived from the weight changes ( top @xmath103-axis ) this line has a slope of @xmath104 . a commercial ( lake shore model drc-91ca ) susceptometer and a homemade one with miniature coils of 2.6 @xmath105 internal diameter were used . the former was convenient for absolute magnitude and a wider range of ac fields while the latter could be used down to 1.3 @xmath20 and gave smoother temperature dependences ( absorption of paramagnetic oxygen can give spurious anomalies in magnetic susceptibility measurements between 40 and 60 @xmath20 ) . they were calibrated by measuring pure lead ( pb ) spheres at low enough frequencies ( 3.3 or 33.3 @xmath106 ) to have negligible eddy current signals in the normal state . the volume ( @xmath107 ) of the powder was found from the weight and x - ray density ( 6.68 mg / mm@xmath108 ) and hence the signal ( @xmath49 ) corresponding to an assembly of perfectly diamagnetic spheres could be found . in our experience the particles settle in the sample capsule so loosely that the magnetic interaction between grains is negligible , but in any case the demagnetization factor of the particles in their holder ( the bottom of a gelatin capsule ) is approximately 1/3 , so the local field acting on any grain will be very close to the applied field @xcite . the london equation @xmath109 is linear , so any diamagnetism arising from current loops that flow out of the @xmath110-plane is additive . since there are two in - plane axes , the appropriate averaging factor for out - of - plane currents and randomly oriented crystallites is 2/3 . so for @xmath111 there is an extra diamagnetic signal of @xmath112 @xmath113 in the measured susceptibility . with @xmath114 @xmath39 and @xmath115 @xmath39 @xcite , this is only @xmath116 @xmath113 on the plots of fig . [ ch05-fig02 ] . the effect would be negligible even if @xmath15 were a factor of 10 smaller . in our experience @xcite @xmath117 often follows the behavior of @xmath118 near room temperature , and this only decreases by a factor of 2 between @xmath119 and @xmath120 @xcite . c. panagopoulos , j. r. cooper , g. b. peacock , i. gameson , p. p. edwards , w. schmidbauer and j. w. hodby , phys . b * 53 * , r2999 ( 1996 ) . c. panagopoulos , j. r. cooper , n. athanassopoulou and j. chrosch , phys . b * 54 * , r12721 ( 1996 ) . d. m. broun _ et al . _ , 99 * , 237003 ( 2007 ) . t. jacobs , s. sridhar , q. li , g. d. gu and n. koshizuka , phys . 75 * , 4516 ( 1995 ) . s. f. lee , d. c. morgan , r. j. ormeno , d. m. broun , r. a. doyle , j. r. waldram and k. kadowaki , phys . * 77 * , 735 ( 1996 ) c. panagopoulos , j. l. tallon and t. xiang , phys . b * 59 * , r6635 ( 1999 ) . y. j. uemura _ et al . _ , . lett . * 62 * , 2317 ( 1989 ) . j. l. tallon , j. w. loram , j. r. cooper , c. panagopoulos and c. bernhard , phys . b * 68 * , 180501(r ) ( 2003 ) . j. w. loram , private communication , june 2009 . y. koike , m. akoshima , i. watanabe and k. nagamine , physica c * 341 - 348 * , 1751 ( 2000 ) . j. r. cooper , l. forr and b. keszei , nature * 343 * , 444 ( 1990 ) . j. r. cooper , h. minami , v. w. wittorff , d. babic and j. w. loram , physica c * 341 - 348 * , 855 - 858 ( 2000 ) . l. forr , phys . lett . a * 179 * , 140 - 144 ( 1993 ) .	we report measurements of ac susceptibility and hence the in - plane london penetration depth on the same samples of bi@xmath0sr@xmath1cacu@xmath2o@xmath3 and bi@xmath4sr@xmath5ca@xmath6y@xmath7cu@xmath2o@xmath3 for many values of the hole concentration ( @xmath8 ) . these support the scenario in which the pseudogap weakens the superconducting response only for @xmath9 .

Subsets and Splits

No saved queries yet

Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.